Pilots and flight attendants work nights, early starts and time zones, and tired crews make mistakes. Airline fatigue risk management is how an airline finds what is making its crews tired and fixes it, from the FAA's duty and rest limits to the fatigue risk management plan every Part 121 airline keeps and the data-driven fatigue risk management system, or FRMS, that some airlines run on top.
This guide is for anyone who runs, joins or pays for that work: fatigue risk managers and safety analysts, directors of flight operations safety, fatigue safety action group members, crew scheduling and pairing planners, pilot and flight attendant union fatigue committees, and the leaders at passenger and cargo airlines, regional carriers and Part 135 operators deciding whether to automate it. No background needed.
Imagine you are a fatigue safety analyst at an airline. On a Sunday morning a first officer files a fatigue report after a red-eye from Las Vegas to Denver: 6 of 7 on the tiredness scale at the top of descent, about 3 hours 40 minutes of broken sleep, and a missed radio frequency change that the captain caught.
Her rest before the flight was 11 hours, more than the rules require. But it began at 1:45 in the afternoon, Las Vegas time.
The roster she flew
Released
1:45 p.m. local
Rest
11 h 00 · legal
Next report
12:45 a.m. local
Lands
4:10 a.m. Denver
Her fatigue report
How tired
6 of 7
Slept
3 h 40
Worst point
Descent
Error
Missed call
Hers is one of 11 reports on the same red-eye in 60 days. Every rule was met. The question is what to change, and whether the change works.
Fatigue risk management answers that question with evidence, before the next tired crew flies the trip.
What airline fatigue risk management is, and what the rules ask of it
The limits, the plans, the systems, the reports, the science and the people, explained without selling anything. If you are new to crew fatigue, start here.
01 · In plain words
Airline fatigue risk management in five minutes
Airline crews work early starts, late finishes, red-eyes and long flights across time zones. Tired people make more mistakes, and accident investigators have named crew fatigue as a likely or contributing factor more than once.1,2 So the FAA sets hard limits for pilots on passenger flights: how long they may be on duty, how long they may fly, and how much rest comes before the next duty.3
Limits are not enough on their own. A rest can be legal and still fall at a time when the body will not sleep. The rule promises a pilot "a minimum of 8 uninterrupted hours of sleep opportunity": an opportunity, not sleep.3 That is why every US airline that flies under Part 121 must also keep a fatigue risk management plan, accepted by the FAA and updated every two years, with a way to check that it works.4,5
Fatigue risk management is that work: collecting fatigue reports, linking each one to the duty that caused it, finding patterns, deciding on a change and checking that it worked. When it runs as a data-driven management system, the industry calls it a fatigue risk management system, or FRMS. In the US an FAA-approved FRMS is also the route Part 117 provides (§117.7) for operating beyond one of its limits.3
The life of one fatigue report, in five steps
A crew member is too tired"Fighting to keep my eyes open" on the descent of a red-eye, after a rest that met every rule.
They report itConfidentially, through a crew app, a safety reporting form or the voluntary program, with how tired and how much sleep.
The report is tied to the dutyThe roster actually flown, the actual flight times, and a fatigue model run on both.
Someone finds the causeOne bad night, or the eleventh report on the same trip? This is where a fix is found, or the report is filed and counted.
A group decides, then measuresThe fatigue safety action group approves a schedule change, and months later someone checks whether it worked.
Steps 3 to 5 are the hard part, and the subject of this guide. The crew member, airline and trip in the example are fictional.
Four things that surprise people
A legal rest can leave a pilot with three hours of sleep
Part 117 requires 10 hours of rest with 8 uninterrupted hours of sleep opportunity before a duty.3 A rest that starts at 1:45 in the afternoon can meet that rule and still leave a pilot unable to sleep before a 1 a.m. report.
Cargo pilots are not under Part 117
The 2012 rule covers passenger flights. All-cargo flights stay under older rules unless the airline opts in.3 The FAA put the cost of covering them at about $452 million against about $3 million in benefits.6 Two 2026 bills to change that sit in committee.7
The plan's two-year clock is written in law
A 2010 act of Congress requires every Part 121 airline to update its pilots' fatigue risk management plan "every 2 years".4 A 2018 act added a plan for flight attendants on the same cycle.5
A fatigue model is not a fatigue system
"A model is not a Fatigue Risk Management System," says IATA's 2025 guidance, and no model should be "the sole go/no-go decision-making tool". Its sleep predictions "may not reflect actual sleep".8
Four words you will see on this page
Sleep opportunity
Time when sleep is possible. The rules protect the opportunity; they cannot promise the sleep.
Window of circadian low
2:00 to 5:59 a.m. body-clock time, when sleepiness peaks. WOCL for short.
Fatigue report
A crew member's confidential account of being too tired, with the duty and the sleep involved.
Fatigue safety action group
The group of managers, crew representatives, schedulers and safety staff that reviews fatigue data and decides on changes.
What is a fatigue risk management system? Limits, plans and systems compared
In shortUS airlines manage crew fatigue three ways at once. Prescriptive limits set the hours. A fatigue risk management plan, which every Part 121 airline must keep, manages fatigue inside those limits. A fatigue risk management system is the data-driven version of the same loop, and in the US the route Part 117 provides (§117.7) to operate beyond a limit.
Question
Prescriptive limits
Fatigue risk management plan (FRMP)
Fatigue risk management system (FRMS)
What it is
Fixed numbers for duty, flight time and rest
An airline's FAA-accepted plan for managing pilot and flight attendant fatigue within the limits
A data-driven management system that can justify operating outside a limit
US basis
14 CFR Part 117 for passenger flights; older Part 121 subparts for all-cargo flights; §121.467 for flight attendants3,9,10
Public Law 111-216 §212 for pilots (2010); Public Law 115-254 §335 for flight attendants (2018)4,5
14 CFR 117.7, and §121.495 for all-cargo operations3,9
Who must have it
Everyone in scope
Every Part 121 air carrier, cargo carriers included
Only an airline that wants to exceed a limit
Approval
The rule itself
FAA review and acceptance; updated every 2 years
FAA approval, showing "at least an equivalent level of safety"
Abroad
ICAO Annex 6 prescriptive regulations; EASA flight time limitations
No direct equivalent; EASA requires fatigue managed inside the management system (Regulation (EU) 965/2012, ORO.FTL.110)
The FAA's definition. A fatigue risk management system is "a management system for a certificate holder to use to mitigate the effects of fatigue in its particular operations. It is a data-driven process and a systematic method used to continuously monitor and manage safety risks associated with fatigue-related error."3
In plain words. A prescriptive rule says "never more than this". Fatigue risk management asks "what is making our people tired, and did our fix work?" Every Part 121 airline already does a version of it: its plan must include "a methodology that continually assesses the effectiveness of the program".4
The loop the rules describe
Collect. Fatigue reports, rosters, actual times and other data, through a reporting system crews trust.3
Identify the hazard. Link each report to the duty behind it, and look across reports for patterns.
Assess the risk. How likely, how severe, using the airline's risk matrix and the science.
Mitigate. Change a pairing, a report time, a hotel or a rest break, "implemented promptly", in EASA's words.12
Assure. Keep "continuous monitoring and regular assessment of the mitigation", and act again if the safety performance is not maintained.12
ICAO groups the same work into four parts: policy and documentation, fatigue risk management processes, safety assurance, and promotion through training and communication.13,11
What investigators have found
Colgan Air 3407 · 2009
"Likely impaired because of fatigue"
The NTSB found the pilots' performance "was likely impaired because of fatigue", but that the extent "cannot be conclusively determined". Fatigue was not in the probable cause. The board also found the airline "did not proactively address the pilot fatigue hazards" of a commuter base.1
UPS 1354 · 2013
Fatigue in the findings
The probable cause was the crew's continued unstabilized approach and failure to monitor altitude. The NTSB found the first officer's fatigue came from "acute sleep loss resulting from her ineffective off-duty time management and circadian factors", and that the captain's performance deficiencies were likely due to factors including fatigue, distraction or confusion. The airline disputed the fatigue finding.2
The pattern
A recurring safety issue
"Reduce fatigue-related accidents" was on the NTSB's Most Wanted List in 2016 through 2020; the NTSB retired the list at the end of 2023.2,14
03 · Part 117 and its limits
Part 117 rest rules and duty limits: the FAA pilot fatigue rules in one place
Status as of October 2026
In shortPart 117 has set flight time, duty and rest limits for pilots on US passenger flights since January 4, 2014. It does not cover all-cargo flights unless the airline opts in, or flight attendants, who have their own rest rule. No new fatigue limits were proposed in 2026.
The FAA published the final rule, "Flightcrew Member Duty and Rest Requirements", on January 4, 2012; a technical correction set its effective date to January 4, 2014.15,16 It applies to flightcrew members and certificate holders "conducting passenger operations under part 121".3
Rule
What it says
§117.5 Fitness for duty
Report "rested and prepared"; no flight duty when "too fatigued to safely perform"; stop when a pilot "has reported him or herself too fatigued to continue"; each pilot "must affirmatively state he or she is fit for duty prior to commencing flight"3
§117.3 Window of circadian low
"A period of maximum sleepiness that occurs between 0200 and 0559 during a physiological night"3
Table A · flight time
With two pilots: 8 hours for a report between midnight and 4:59 a.m. or between 8 and 11:59 p.m.; 9 hours from 5 a.m. to 7:59 p.m.3
§117.11 · augmented crews
13 hours of flight time with three pilots; 17 hours with four3
Table B · flight duty period
From 9 hours for a start between midnight and 3:59 a.m., up to 14 hours for a 7 to 11:59 a.m. start with one or two flights; less for more flights3
Table C · augmented duty
With four pilots and the best rest facility, 17 to 19 hours depending on the start time; 19 hours only for starts from 7 a.m. to 12:59 p.m.3
§117.19 Extensions
Up to 2 hours for unforeseen events before takeoff; more than 30 minutes only once before a 30-hour rest; report extensions over 30 minutes to the FAA within 10 days3
§117.23 Cumulative limits
Flight time: 100 hours in any 672 consecutive hours, 1,000 in any 365 days. Duty: 60 hours in any 168, 190 in any 6723
§117.25(b) Weekly rest
30 consecutive hours free from all duty in the past 1683
§117.25(e) Rest before duty
"At least 10 consecutive hours" before a flight duty period or reserve, which "must provide … a minimum of 8 uninterrupted hours of sleep opportunity"3
§117.25(f) Speak up
A pilot whose rest "will not provide eight uninterrupted hours of sleep opportunity" must tell the airline, and cannot report until given a compliant rest3
§117.9 Training
Annual fatigue education and awareness training for pilots, dispatchers, schedulers, operational control staff and their managers; the program updated every two years3
Shortened from 14 CFR Part 117 as of October 2026.3 Reserve rules (§117.21), unacclimated pilots (a 30-minute reduction) and consecutive night duties (§117.27) add more detail. Read the regulation before relying on any single number.
Check a rest period against two Part 117 definitions
The rest and the duty after it
0 h 00rest period
–at least 10 hours (§117.25(e))
–includes 1 to 7 a.m., a "physiological night's rest" (§117.3)
0 h 00of the next duty inside 2:00 to 5:59 a.m., the window of circadian low
Body-clock time is the time zone the crew member is acclimated to. The starting numbers are the fictional red-eye from the example, in Denver time. This checks two definitions in §117.3 and the 10-hour minimum; it is not a legality check of a whole duty, and it cannot say whether anyone slept.3
Who Part 117 does not cover
All-cargo flights: the carve-out stands. All-cargo operations stay under Part 121 Subparts Q, R and S unless they elect Part 117.3,9 In 2014 the FAA estimated that applying the rule to them would cost about $452 million against base-case benefits of about $3 million (about $10 million in a high case), and made no change.6 As of October 2026, H.R. 7191 and H.R. 7526 would extend Part 117 to all-cargo flights; both sit in the House Transportation and Infrastructure Committee.7 Cargo carriers are still Part 121 air carriers, so the fatigue risk management plan duty applies to them.4
Flight attendants: their own rest rule. Since the January 10, 2023 compliance date, a flight attendant scheduled for a duty period of 14 hours or less must get at least 10 consecutive hours of rest, which "may not be reduced".10,17 Longer duties need at least 12 hours, reducible to 10 only if a 14-hour rest follows.10 So a pairing change that is legal for the pilots must also be checked against §121.467 and the flight attendant contract.
The FAA's July 2026 proposal on duty and rest is about preempting state meal and rest-break laws; comments closed September 4, 2026. It does not propose new fatigue limits.18 We found no other FAA fatigue rulemaking in 2025 or 2026.
04 · An FRMS under §117.7
FRMS under 14 CFR 117.7: what an FAA approval involves
Status as of October 2026
In shortAn airline that wants to exceed any Part 117 limit, for an ultra-long-haul route for example, needs an FAA-approved fatigue risk management system that gives at least an equivalent level of safety. It must have six elements and collect data specific to the operation it wants to fly.
No certificate holder may exceed any provision of this part unless approved by the FAA under a Fatigue Risk Management System that provides at least an equivalent level of safety against fatigue-related accidents or incidents as the other provisions of this part.14 CFR 117.7(a)3
The six elements of an FRMS
1 · Policy
A fatigue risk management policy
2 · Training
An education and awareness training program
3 · Reporting
A fatigue reporting system
4 · Monitoring
A system for monitoring flightcrew fatigue
5 · Incidents
An incident reporting process
6 · Evaluation
A performance evaluation
14 CFR 117.7(b).3 The all-cargo rules list the same six elements for an airline that wants to exceed their limits (§121.495).9
The guidance behind it
Document
Subject
Date and status
AC 120-103A
Fatigue risk management systems for aviation safety: an optional alternative to prescriptive limits that must show an equivalent level of safety
Data, specific to the operation. The FAA's 2026 renewal of the Part 117 paperwork approval says airlines that choose an FRMS "are required to collect data specific to the need of the operation for which they will seek an FRMS authorization". It counts 52 air carriers and an estimated 30,954 burden hours a year for Part 117 records.21 Those are paperwork estimates, not a measure of the whole job.
Ultra-long-haul. With four pilots, flight time is capped at 17 hours, and Table C allows a duty of 19 hours only for certain start times.3 A route that needs more cannot be scheduled under Part 117 alone; it needs an approved FRMS and the data to support it. We found no public count of US airlines holding a §117.7 approval.
05 · The fatigue risk management plan
The fatigue risk management plan (FRMP): what the law requires, and the two-year clock
Status as of October 2026
In shortEvery Part 121 airline must keep an FAA-accepted fatigue risk management plan for its pilots, under a 2010 law, and one for its flight attendants, under a 2018 law. Each must be updated every two years, include annual fatigue training, and show a way to check that the plan works.
Pilots · Public Law 111-216 §212 · August 1, 2010
Each Part 121 air carrier submits "a fatigue risk management plan for the carrier's pilots", originally within 90 days4
Contents: current flight time and duty limits; "a rest scheme consistent with such limitations that enables the management of pilot fatigue", including annual training on fatigue, its effects and countermeasures4
"A methodology that continually assesses the effectiveness of the program", including its ability "to improve alertness" and "to mitigate performance errors"4
Updated "every 2 years" and submitted for review and acceptance; the FAA reviews an update within 12 months4
The airline "shall comply with" its accepted plan; violations carry civil penalties4
Flight attendants · Public Law 115-254 §335 · October 5, 2018
The same duty, "for the carrier's flight attendants": submit within 90 days, update "every 2 years", comply, with the same civil-penalty clause5
The same act set the 10-hour flight attendant rest that became §121.4675,10
The FAA's 2026 paperwork notice for this plan counts 55 Part 121 carriers and 2 new entrants, and estimates 20 hours for an initial plan and 5 hours for each update22
The dates
The guidance, and a common mix-up
FAA guidance on the plan. The FAA's information notices for operators cover the plan: InFO 10013 and 10017 (August 2010) on writing it, InFO 13008 on required updates, and InFO 19007 on the flight attendant plan.23 FAA guidance expresses the update cycle as 24 calendar months, tied to the airline's operations specifications.23
Cite the law, not a training notice. The two-year update comes from the statutes: §212 for pilots and §335 for flight attendants.4,5 InFO 15001 (January 2015) is about the §117.9 fatigue education and awareness training, not the plan's update cycle.23 Training is annual; the training program, like the plan, is updated every two years.3
Penalties for breaking the plan duty follow the general civil-penalty statute and its inflation adjustments; we quote no dollar figure.4,24 We found no public FAA enforcement case specific to a fatigue risk management plan or Part 117 as of October 2026.
06 · Safety management systems
Safety management systems (Part 5): where fatigue fits, and the 2027 date for Part 135
Status as of October 2026
In shortFatigue is one hazard inside an airline's safety management system. Part 121 airlines have had to run an SMS since the 2015 rule and revise it to the 2024 rule by May 28, 2025. Part 135 operators and air tour operators must have one by May 28, 2027, including a confidential reporting system where employees can report hazards without fear of reprisal.
When
What happens
January 8, 2015
Part 5 rule published, requiring an SMS for Part 121 airlines25
April 26, 2024
Final rule "Safety Management Systems" published, effective May 28, 2024; extends SMS to Part 135, §91.147 air tour operators and certain design and production approval holders25
May 28, 2025
Part 121 airlines with an existing SMS must have revised it to the new Part 5 (§5.7)25
May 28, 2027
Part 135 certificate holders and §91.147 operators must implement an SMS and submit a declaration of compliance (§5.9)25,26
We found no extension of these dates in the Federal Register or FAA notices as of October 2026.
What Part 5 asks that touches fatigue
§5.21 · Policy
Reporting, and its limits
A safety policy that covers "employee reporting of safety hazards or issues" and defines "unacceptable behavior and conditions for disciplinary action".25
§5.25 · Accountability
One accountable executive
An accountable executive who is "the final authority over operations" and is accountable for the SMS.25
§5.71 · Assurance
A confidential channel
"A confidential employee reporting system in which employees can report hazards … without concern of reprisal for reporting."25 Fatigue reports are one stream of it.
How the pieces fit. The SMS has four parts: safety policy, safety risk management, safety assurance and safety promotion (§5.5).25 Fatigue is one hazard within it. The fatigue risk management plan, and any FRMS, are the specialized fatigue part: fatigue reports feed the SMS's hazard identification, and measured fixes feed its safety assurance.
For Part 135 operators. Part 117 and the plan statute do not apply to Part 135; it has its own duty and rest rules. But the May 28, 2027 SMS date, with its confidential reporting system, is a near-term reason to formalize how fatigue is reported, reviewed and acted on.25
07 · Fatigue reports and just culture
Pilot fatigue reports, fatigue calls and just culture
In shortThe rules expect crews to speak up when they are too tired, and require a fatigue reporting system in any FRMS. Reports only keep coming when crews trust they are confidential and not used to punish. The tiredness scale, the sleep and the duty in each report are the raw material of everything else.
Where the rules ask crews to speak up
§117.5(c)
Too fatigued to continue
No airline may let a pilot continue a duty once the pilot "has reported him or herself too fatigued to continue".3
§117.5(d)
Fit for duty, out loud
Each pilot "must affirmatively state he or she is fit for duty prior to commencing flight".3
§117.25(f)
A rest that will not work
A pilot whose rest will not give "eight uninterrupted hours of sleep opportunity" must tell the airline before reporting.3
What a good fatigue report captures
The Samn-Perelli scale · 7 points
Fully alert, wide awake
Very lively
Okay, somewhat fresh
A little tired
Moderately tired
Extremely tired
Completely exhausted
The 9-point Karolinska Sleepiness Scale, from "extremely alert" to "extremely sleepy, fighting sleep", is also common.27 Airlines often add a line on what the rating means, such as "very hard to concentrate".
The rest of the report
The flight, the pairing and the day of the trip
When the crew member felt worst: cruise, descent, a ground delay
Sleep in the 24 hours before the report, and its quality
What contributed: layover timing, the hotel room, an early report, commuting
Any error or near miss, and whether they continued
In their own words, which often carry the cause
Confidential, and still accountable
Voluntary safety reporting. Many airlines also receive fatigue reports through ASAP, the Aviation Safety Action Program: voluntary reporting under an agreement between the airline, the labor group and the FAA. Its advisory circular, AC 120-66C (March 31, 2020), frames it as resolving safety issues "through corrective action rather than punishment".28 A report that is both a fatigue report and an ASAP report has to be handled under the terms of that agreement.
Just culture. Honest reports are learned from; reckless behavior is still addressed. Part 5 asks for both: a confidential, no-reprisal reporting system, and a policy that defines what is unacceptable.25 Pilot associations ask that the fatigue safety action group include crew representatives, who guard the confidentiality of the reporting system.29
Fatigue calls. In airline usage a fatigue call is a crew member removing themselves from a trip as too fatigued to fly safely; the duty to stop is in §117.5.3 How a fatigue call is paid and reviewed is set by each airline and its labor contracts, not by an FAA rule. A sudden drop in fatigue reports can mean crews stopped reporting, so fatigue calls are worth watching alongside them. We found no public study that measures how much confidentiality changes report volume; treat it as a principle, not a statistic.
08 · Fatigue models and their limits
Biomathematical fatigue models: what they do, and what they cannot
Status as of October 2026
In shortA biomathematical fatigue model estimates alertness across a schedule from time awake, time of day, sleep opportunity and time zones. It is a useful way to compare schedules and find fatigue drivers. It is not an FRMS, it describes an average person, and its sleep is an assumption unless someone tells it otherwise.
Models typically turn a roster into a score on a scale, often 0 to 100, for each duty, and highlight where the score dips. They come as standalone schedule-analysis tools, as alertness models built into crew pairing and rostering optimizers, as day-of-operations scoring through an interface, inside crew apps that log sleep, and with consulting.
What IATA's 2025 guidance says
"Biomathematical models (BMM) are an optional tool for fatigue hazard identification." "A model is not a Fatigue Risk Management System (FRMS)." "An optimizer used in conjunction with a model does not constitute an FRMS."8
"A model should not be used by individuals, operators or regulators as the sole go/no-go decision-making tool." "The output of a model does not necessarily correlate to a safety risk."8
"Model outputs represent the population average and may not be accurate for specific individuals." "Models may not consider the operational context and mitigations."8
"Model sleep predictions may not reflect actual sleep duration and quality which are fundamental to the validity of a model's output."8
A model "can be a useful tool to analyze the potential fatigue exposure and identify fatigue drivers in a safety event, if combined with crew member's perceptions and sleep history."8
IATA Human Factors Task Force, "Uses and Limitations of Biomathematical Fatigue Models", April 2025.
Planned and flown rosters differ. A 2025 peer-reviewed study at one airline ran a model over both: 157 fatigue reports in 78,061 planned duties, and 235 in 82,612 duties as completed. The model's odds ratio for flagging the duties that drew reports was 3.04 on the planned schedules but 0.41 on the completed ones; completed duties started later and were shorter.30 Which roster a model reads changes what it finds.
Check the validation for your operation. Validation claims usually come from the model's maker. Ask what evidence exists for operations like yours, whether a tool says when a duty falls outside the range the model was validated for, and how its thresholds were set.8
From that study we derive about 2.0 reports per 1,000 planned duties and 2.8 per 1,000 completed duties at that one airline.30 It is one airline, not a benchmark. We found no public industry-wide fatigue report rate.
09 · ICAO and EASA
The international picture: ICAO and EASA fatigue risk management
Status as of October 2026
In shortICAO made FRMS an international standard in 2011 as an alternative to prescriptive limits. Europe requires every operator to manage fatigue inside its management system, and a full fatigue risk management process when it departs from the standard scheme. A Commission regulation adopted September 8, 2026, not confirmed in force as of October 2026, would name the fatigue safety action group in EU law.
ICAO
ICAO's Council adopted FRMS Standards for Annex 6 in June 2011 as "an alternative to prescriptive flight and duty limitations", applicable from December 15, 201131
States may use prescriptive limits, FRMS for all operations, or FRMS for some and prescriptive limits for the rest11
Doc 9966, "Manual for the Oversight of Fatigue Management Approaches": current English text Second Edition, Version 2 (Revised), 202013
The Fatigue Management Guide for Airline Operators, second edition, 2015, from ICAO, IATA and IFALPA13,32
EASA
Flight time limitations for commercial air transport sit in Regulation (EU) 965/2012, Subpart ORO.FTL33,34
ORO.FTL.110 makes fatigue a managed hazard for every operator; ORO.FTL.120 requires a full fatigue risk management process "when FRM is required", for example when an operator departs from the prescriptive scheme12,34
The operator "shall take mitigating actions when the FRM safety assurance process shows that the required safety performance is not maintained"12
Commission Delegated Regulation C(2026) 6142 of September 8, 2026 would replace ORO.FTL.120 with "Fatigue risk management system (FRMS)" and require "a fatigue safety action group (FSAG)"; its entry into force was not confirmed as of October 202633,35
For a US airline the FAA's rules govern; ICAO and EASA matter for international operations, codeshare partners and as a reference for what a mature fatigue risk management system looks like.
10 · How airlines do it today
How airlines manage crew fatigue today, and who is involved
In shortMany airlines combine a small fatigue team, a safety reporting tool, a fatigue model and a fatigue safety action group. The hard part is the linking: tying each report to the roster actually flown, the actual times and a model run, then seeing the pattern across reports in time to act.
1 · In-house
The fatigue team
A fatigue risk manager and analysts read reports from the reporting tool, pull rosters and times by hand, and prepare cases for the fatigue safety action group.
Strength
Knows the operation, the crews and the contracts
Strain
Linking can take hours per report (no published figure), so often only the serious ones are worked
2 · Science
Fatigue model tools
Schedule-analysis tools score pairings and rosters; some can read rosters as flown and take reports as markers on a duty.
Crew pairing and rostering optimizers can treat predicted fatigue as a cost or a rule when they build schedules.
Strength
Fatigue designed out before the schedule is published
Watch for
"An optimizer used in conjunction with a model does not constitute an FRMS"8
4 · Safety tools
Reporting platforms and consultants
SMS and safety reporting platforms add fatigue forms, queues and risk matrices; consultants design programs, audit them and analyze incidents.
Strength
Reports captured and tracked; expert judgment
Watch for
Reports filed and counted, not linked to the duty
A manual case and an assisted one, side by side
Question
Typical manual process
Process assisted by software (what to look for)
Which reports are worked?
The ones that look serious; the rest are acknowledged and counted
Every report read and linked the day it arrives
The roster?
Pulled by hand from crew scheduling, often the planned version
The days before the report as actually flown, with actual out, off, on and in times
The model?
Re-keyed into a separate tool, with the model's assumed sleep
Run on the roster flown, with assumed sleep and with the sleep the crew member reported
Patterns?
Counted by month and fleet; found at the periodic review
Grouped by pairing, flight, hotel and hour, against the airline's own rate, as reports arrive
Fixes?
Proposed to scheduling, then checked for legality and cost
Tested for alertness, Part 117, the labor contract and cost before they are proposed
Did it work?
Checked when someone remembers
Measured for a set period, with fatigue calls watched as well as reports
The time a manual case takes is not published anywhere we could find; one working estimate is an afternoon to link a single report, and it is only an estimate. IATA's guidance on safety performance indicators suggests combining roster, reporting, survey, performance, sleep and model measures, reviewed by the fatigue safety action group.32
Who is involved, and what each cares about
FR
Fatigue risk manager or FRMS manager
Reads reports, decides which become cases, prepares the fatigue safety action group, and owns the plan and the FAA relationship on fatigue. Cares about which pairing is making people tired, proof the fix worked, and confidentiality.
SA
Safety analyst
Works trends, patterns and the evidence for the FAA. Cares about seeing patterns before the periodic review, and numbers that trace to their source.
DS
Director of flight operations safety
Owns the SMS and the fatigue program, and often chairs the fatigue safety action group. Cares about evidence that reports are acted on, and records an inspector can follow.
FS
Fatigue safety action group
Flight operations management, pilot and flight attendant representatives, scheduling, training and safety, sometimes a sleep scientist.29 Cares about evidence, options and cost on one page, and a recorded vote.
CP
Crew scheduling and pairing planners
Build pairings and rosters and put approved changes into the next bid period. Care about a specific change that is legal under Part 117, §121.467 and the contract, with a date.
UC
Union fatigue and safety committees
Represent pilots and flight attendants, guard confidentiality and often co-own the voluntary reporting agreement. Care that no reporter is exposed and that reports lead to fixes.
AE
Accountable executive and the VP who funds crews
Final authority over operations under Part 5, and the budget for reserves, hotels and crew days.25 Care about fatigue calls, reliability and risk accepted at the right level.
PI
FAA principal operations inspector
Oversees the plan and any FRMS. Cares about updates on time, training records and data showing the plan works.
Part B · See it working
Botminds Fatigue Risk Management System, one fatigue report at a time
Part A described the work without naming a product. This part shows how Botminds does it, at Corvane Airways, a fictional US network airline with crew bases across the US. Every name, report, pairing, hotel and number in the screens is sample data.
11 · One report, start to finish
Follow one red-eye from a fatigue report to a signed board decision
In shortThis is the example from the top of the page, inside the product. Agents read the report, link the roster actually flown, run the fatigue model twice, find the pattern and test pairing changes. Megan Ashford, the fatigue risk manager, decides and signs; the review board votes; crew planning publishes. The report arrives on Sunday morning, and on Tuesday morning the case goes to Thursday's board.
1
The fatigue report is read the moment it arrives
Report R-26-1187 comes in from the crew app on Sunday, October 18 at 6:52 a.m., Denver time, after red-eye 2290 from Las Vegas to Denver. Within a minute every field is read, each with a confidence: the flight, the pairing, when she felt worst, how tired (6 of 7), her sleep (3 h 40, broken) and what contributed, in her words: an afternoon layover, a room facing the pool deck, her third time on this red-eye this month.
Her name is sealed. She appears as reporter R-7731, and only the fatigue risk manager can reveal her, with a reason that is logged.
The report, read. Each field the case relies on, with its confidence and a link back to the words it came from.
2
Her words, the roster she flew and her alertness, on one time axis
The roster as actually flown is linked from crew scheduling and operations: released at 2:45 p.m. Denver time (1:45 p.m. in Las Vegas), an 11-hour rest that meets §117.25(e), the sleep she reported from 5:10 to 8:50 p.m., then a 1:45 a.m. report and a flight that lands at 4:10 a.m., inside the window of circadian low.
The fatigue model runs twice. With the layover sleep the model assumes, 6 h 10, her alertness at landing is 66. With the 3 h 40 she reported, it is 58, below the airline's high-risk line of 62.
One time axis. In her words, the roster actually flown, and alertness with the sleep she reported against the sleep the model assumed. The body-clock low is shaded.
3
Eleven reports on one pairing, and four fixes tested
The pattern was raised on Sunday at 7:04 a.m.: 11 fatigue reports on this red-eye in 60 days, 9 pilot reports and 2 flight attendant reports, plus 2 fatigue calls. Nine pilot reports in 120 pilot duties is 75 per 1,000, against 4.1 for the airline: 18 times the rate. Small numbers, so it is read as a strong signal, not a statistic.
Four pairing changes were scored for alertness, Part 117, the pilot contract and cost. Option A, arriving the evening before for a full night in bed, lifts alertness at landing from 58 to 74 for an estimated $78,110 a year. Option C scores best but moves a revenue flight, so it goes to network planning as a note. Option D breaks the 10-hour rest rule and is refused, with the rule named.
The options. Each with its alertness curve, Part 117 and contract checks, and cost a year. Costs are estimates in sample data.
4
Sent to the board, signed
A case writer drafted the hazard, an afternoon layover before a red-eye that lands in the body-clock low, rated it 4C, high, and recommended option A. A case checker re-opened all 23 sources: 22 confirmed, and one claim softened to "9 of 11 reports give 3 h 40 to 4 h 55".
Megan asks why option A over option B, reads the cited answer, chooses reason code C2, "pattern confirmed, schedule change needed", and signs. On Thursday three of five voting members, crew representatives among them, approve; the crew pairing planner publishes option A for the December 1 bid period, and the change is watched for 90 days.
The decision. Hazard, recommendation and identity on one card, a reason code, and an electronic signature with the name and time.
12 · Every screen
The rest of the work: the morning, the inbox, patterns, the board, assurance and records
In shortOne report is the story. These screens show the whole desk: what needs a person each morning, the confidential inbox, the crew app, patterns across reports, the review board, whether changes worked, the plan and training records, the audit trail and the airline's own rules.
Today. What was read overnight, the open cases, the next board and the plan's update date, on one line.
Needs you. The highest-risk case leads, with the reason in one paragraph and every number cited.
How it was built. Each step with its time and what it found: Sunday 06:53–07:04, Tuesday 06:30–07:06. Every step only read; nothing changed a roster.
The confidential inbox. Reporters appear as codes. Reports that need a person, such as an illness that may not be fatigue or a duplicate, are flagged with a proposed path.
The crew app. The form, shown on a later report: a few taps and a few words, with the promise that the name is sealed.
A pattern. When crews felt it, by hour, and the rate against the airline's own. A person decides: draft a case, watch, or dismiss with a reason.
Identity, sealed. Questions to the reporter go through the confidential channel; a reveal needs a reason and is logged.
The rate over a year. A falling rate with steady reporting is the goal; fatigue calls are watched too.
The review board. One decision pack per case, de-identified, with each member's vote and reason code. Many airlines call this group the fatigue safety action group.
Is it working? Every change is measured for 90 days, and a case closes only when a person says it worked.
The plan. The two-year update, with the changes drafted from the year's cases for a person to approve.
Training. Annual fatigue training due in the next 30 days, by group: §117.9 for pilots, dispatchers, schedulers and managers; the flight attendant plan for cabin crew.
Audit trail. Every action, by people and agents, chained so that a change to any line shows.
Roles. Who may read, reveal, send to the board, vote, publish and change settings. No agent can reveal, send, vote or publish.
Thresholds and rules. The airline's own lines and pattern rule; a change needs a reason and is logged.
Systems of record. What it reads, and how often.
Select any screen to open it full size. The airline, people, reports, pairings and hotels are fictional; figures in the screens are sample data.
13 · Agents, and who signs
Eight agents do the reading, linking and arithmetic. People make the decisions.
In shortEach agent has one job and one thing it never does. None of them reveals a reporter, changes a roster, contacts a crew member outside the confidential channel or closes a case. The fatigue risk manager signs every case to the board, the board votes, and only a crew planner publishes a change.
The eight agents
Report reader
Reads each report from the crew app, email, a scanned form or a cross-filed safety report; extracts the flight, duty, sleep, rating and factors; seals the name.
Never shows a reporter's identity to anyone but the gatekeeper.
Roster linker
Pulls the 7 days of roster before the report and the actual out, off, on and in times; computes duty, rest and time in the body-clock low.
Never changes a roster.
Fatigue model
Runs the airline's fatigue model on the roster flown, with planned sleep and with reported sleep, and explains the difference.
Never treats a score as a verdict, and says when a duty is outside the model's validated range.
Pattern finder
Groups reports by pairing, flight, hotel, station, hour and fleet, and compares each group with the airline's own rate.
Never opens a case by itself; a person decides.
Case writer
Drafts the case: hazard, evidence, risk rating on the airline's matrix, and options with their predicted effect.
Never sends a case to the board.
Case checker
Re-opens every source the draft cites, challenges each claim, and marks what it could not verify.
Never approves a case.
Pairing planner
Builds options with the pairing optimizer and scores each for alertness, Part 117 legality, the labor contract and cost.
Never publishes a pairing; a change to a revenue flight goes to network planning as a note.
Assurance monitor
Measures each change for 90 days, tracks the plan and training clocks, and builds evidence packs for the FAA.
Never closes a case; packs need a person's signature.
People decide; everything is cited
Decides and signsFatigue risk managerApproves or edits each draft, sends schedule changes to the board with a reason code and signature, closes cases with a reason. The only role that may reveal a reporter.
VotesFRMS review boardVotes in the product with reason codes; the chair signs the minutes. In the screens the quorum is 3 of 5 voting members with a crew representative, this airline's policy.
PublishesCrew pairing plannerPublishes only a change the board approved, in the next bid period.
Every value on a case cites its source: the report text, the roster, the actual times, the model run or the rule. Agent statements and human decisions are kept apart on the audit trail, and board packs and exports are de-identified.
Reads every new report, decides which become cases, prepares the board and owns the plan.
"I don't need more reports. I need to know which pairing is making people tired, and proof that the fix worked."
VR
Capt. Victor Ruiz · director of flight operations safety, board chair
Chairs the monthly board and signs decisions and minutes.
"Bring me the evidence, the options and the cost on one page, and I'll get a decision in one meeting."
LO
Capt. Lena Ostrow · board member, pilots' fatigue committee
Votes on cases and protects the confidentiality of the reporting system.
"If a pilot thinks a report can come back to them, the reports stop. That is the whole system gone."
JP
Jordan Pike · senior crew pairing planner
Builds the monthly pairings and puts approved changes into the next bid period.
"Tell me which pairing, what to change and by when. Don't hand me a change that breaks Part 117 or the contract."
HR
Hannah Reyes · safety analyst
Works patterns, trends and the evidence for the FAA.
"I want the pattern before the quarterly review, not in it."
CD
Colin Draper · FRMS administrator
Connects systems and sets thresholds, roles and reason codes.
"One place to set who can see what, and a log that shows it held."
14 · What goes in and out
Fatigue reports in, a measured decision out, with a person and a source in between
In shortIt reads the reports and the systems around them, writes nothing back to a roster, and leaves behind linked and cited cases, de-identified board packs, signed decisions, measured results and the records the FAA expects.
The reports
Crew app fatigue reports
Email and scanned paper forms
Cross-filed voluntary safety reports
Fatigue calls
Around them
Rosters actually flown
Out, off, on and in times
Pairings and the optimizer's options
Your fatigue model
Hotels, contracts and training records
Your plan, risk matrix and reason codes
Agents read, link, model, draftEvery report, against the night actually flown, under your rules and Part 117
A person signs each decision
Audit trail on every step
Outputs
Each report linked to its duty, the same day
Two model runs, planned and reported sleep
Patterns against your own rate
Cases with cited evidence and tested options
De-identified board packs, votes and minutes
Before-and-after results for every change
Plan updates, training status and FAA evidence packs
What changes for the team
Every report linked, not only the serious ones
Cases to edit, not to assemble
Patterns before the periodic review
Systems to plan for
Crew schedulingRosters actually flown
Crew pairing optimizerPairings and options
Operations dataOut, off, on and in times
Crew app or reporting formReports and the confidential follow-up
Safety reporting and SMSCross-filed reports, kept apart
Fatigue modelRuns on actual and planned rosters
Hotel bookingsLayover hotels and rooms
Training systemAnnual fatigue training
For each, check whether it has an interface, a file feed, or only screens, and who owns it. Identity and access, and email or chat for notices, complete the list.
15 · With what you already run
FRMS software that works alongside your scheduling, your fatigue model, your SMS and your unions
In shortKeep your crew scheduling system, your pairing optimizer, your fatigue model, your safety reporting system and your fatigue safety action group. The new piece is the linking and the follow-through: every report tied to the night actually flown, the pattern found as it forms, and every change measured.
Your crew scheduling and pairing systems
They stay the systems of record. Rosters and options are read from them; an approved change is published by your planners, in your next bid period.
Your fatigue model
Keep the model you license. It runs on the roster actually flown, with the sleep crews report as well as the sleep it assumes.
Your safety reporting system and SMS
Fatigue reports that are also voluntary safety reports stay under that program's terms; only de-identified facts are shared.
Your union fatigue committees
Crew representatives sit on the board and vote in the product; every identity reveal is logged and reported to the fatigue committee.
Your fatigue safety action group
It meets as it does today, with one de-identified decision pack per case, recorded votes and drafted minutes.
Your FAA relationship
The plan update, training status and evidence that reports are acted on, each line traced to its source.
Three principles
1
Read every report against the night actually flown
The roster flown, the actual times and the sleep the crew member reported, not only the plan and an assumed sleep.
2
Seal the reporter
Identity is protected by role, not by habit. One person may reveal it, with a reason, and every reveal is logged.
3
Agents find; people decide
Agents read, link, model and draft. The fatigue risk manager signs, the board votes, a planner publishes, and every step is on the audit trail.
Built on Botminds
One platform for agents, data and screens
Builder: describe the work, get a spec, build it, test it.
Runtime: agents, data, integrations, screens and models, in one place.
SecureIsolatedGovernedTestedScalableHighly available
Enterprise-ready
Roles and separation of duties; one gatekeeper for identity
Reason codes and electronic signatures on every decision
An audit trail and evidence packs an inspector can read
Integrations with scheduling, operations, safety reporting and your fatigue model
Human approval and citations everywhere
Part C · Are you ready to automate?
Should you automate fatigue risk management? A buyer's view
You run fatigue risk management, sit on the group that decides, or pay for the reserves when crews call in fatigued. This part is written for you: what fatigue and its follow-up cost today, what automating the linking and follow-through could return, whether your team is ready, and what to ask any vendor of FRMS software, including us.
16 · What fatigue costs you
What crew fatigue and its paperwork cost, before you buy anything
In shortStart from your own numbers. We found no public figure, as of October 2026, for the cost of a fatigue call, the time to work a fatigue report, or the cost of running a fatigue program, so the figures that matter are yours. Once you trust them, you can judge any solution against them, ours included.
30,954
hours a year: the FAA's own estimate of the paperwork burden of Part 117 records, including FRMS data and training records, across 52 air carriers.21
FAA information collection 2120-0751, 2026 · paperwork only
2 yrs
the longest a Part 121 fatigue risk management plan may go without an update, for pilots and for flight attendants. Training is annual.4,5,3
Public Law 111-216 §212 · Public Law 115-254 §335
2–3
fatigue reports per 1,000 duties at the one airline in a 2025 peer-reviewed study, derived from its counts. Not an industry benchmark.30
Derived from Devine and others, Safety, 2025
The FAA estimates 20 hours to write an initial flight attendant plan and 5 hours for each update.22 These official figures count paperwork only; they leave out reading reports, linking rosters, running models, preparing the group and measuring results.
What nobody publishes. We found no public figure for the time to work a fatigue report by hand, the cost of a fatigue call, the cost of running a plan or an FRMS, an industry-wide fatigue report rate, the share of reports that lead to a schedule change, the number of US airlines with a §117.7 approval, or the market size for FRMS software. Vendor figures we found came without a dataset. Measure your own.
Find your own numbers
Cost
Where it hides
How to find your number in a week
Analyst time
Pulling rosters and times, re-keying the model, finding similar reports
Time five reports worked in depth; multiply by the reports you would work if you could
Last year's fatigue calls from crew scheduling, and what operations and finance say one costs
Reports not worked
Acknowledged, filed and counted
Share of last quarter's reports that were linked to a roster and a model run
Late patterns
Clusters found at the periodic review, months after the first report
For your last three schedule changes, days from the first report to the decision
Changes not measured
Fixes approved, never checked
Share of last year's changes with a before-and-after result on file
Records and the FAA
Plan updates, training tracking, evidence for the inspector
Hours spent on the last plan update and the last FAA request
Costs no calculator counts
An error on a tired nightThe missed call the captain caught, and the one nobody catches.
Trust in reportingOne exposed reporter, and the reports stop.
The plan's credibilityThe plan must show a way to check it works.4
Crew goodwillReports that never change anything teach crews not to report.
Your best peopleAnalysts spend their time linking instead of judging.
17 · Your return
What automating fatigue risk management could return for you
In shortPut in your own numbers. There are no published benchmarks for any of these inputs, so every default below is marked illustrative: an assumption to replace, not a benchmark or a promise.
The value levers
Analyst time
Cases to edit, not to assemble
Rosters, times, model runs and similar reports arrive linked; the analyst starts from the evidence.
reports worked × hours saved × hourly cost
Depth
Every report linked
Not only the reports that look serious: the quiet ones are where a pattern starts.
share of reports linked, before and after
Fatigue calls
Fixes that reach the pairing
A confirmed change on the pairing that drives calls, measured afterwards.
calls × cost × share reached × fall in calls
Speed
Patterns as they form
A cluster raised the day it crosses your rule, not at the quarterly review.
days from first report to decision
Assurance
Every change measured
Before and after for each change, with fatigue calls watched beside reports.
share of changes with a measured result
Records
Evidence ready to hand
The plan update, training status and evidence for the inspector, each line traced.
hours a year assembling records × share saved
The calculator
Every input marked illustrative is an assumption, not a published figure. Replace each one with your own number.
Fatigue reportsFatigue callsRecords
Illustrative yearly value
Analyst time on reports today, valued$0
Analyst time saved on the same reports$0
Fatigue-call cost avoided$0
Records time saved$0
Total$0
Analyst hours freed a year0
Cost to work every report in depth, with the linking done for you$0
Formulas, per year: Time today = reports × share worked × hours by hand × hourly cost Time saved = reports × share worked × (hours by hand − hours with linking) × hourly cost Calls avoided = calls × cost per call × share reached × fall in calls Records saved = records hours × share saved × hourly cost Every report in depth = reports × hours with linking × hourly cost
Not counted: errors avoided, trust in reporting, patterns found earlier, and changes that are measured instead of assumed. Weigh the result against the cost of the solution and your rollout.
Start here18 · Readiness check
Are you ready to automate fatigue risk management? Nine questions, five minutes
In shortPick the answer closest to your team today. Eight questions score 0 to 2; the last is for reflection. The meter adds up your score as you go, and each question says why it matters and what a strong answer sounds like.
1
Volume
About how many fatigue reports does your airline receive in a year, pilots and flight attendants together?
Why it mattersVolume turns hours per report into analysts, and decides whether every report can be worked in depth.
A strong answer"About 1,500 a year, rising since we launched the crew app."
2
How reports are worked today
How many reports are linked to the roster actually flown and a model run?
Why it mattersModel sleep "may not reflect actual sleep".8 Where only the serious reports are linked, that gap is where automation pays off most.
A strong answer"The serious ones, by hand; the rest are acknowledged and counted."
3
Your baseline
Do you know your hours per report, your report rate per 1,000 duties and your days from first report to decision?
Why it mattersNo public benchmark exists for any of them, so your own baseline is the only way to prove an improvement.
A strong answer"About four hours a report, 4 per 1,000 duties, and two months to a schedule change."
4
Pressure
Is there a date that matters: a plan update due, an FAA question, an ultra-long-haul route, the 2027 SMS date for Part 135?
Why it mattersA real date turns interest into a plan. Each plan must be updated every two years.4,5
A strong answer"Our plan update is due in the spring, and we want the evidence ready before then."
5
Data and systems
Can software reach the roster actually flown, the actual times, your reports and your fatigue model?
Why it mattersLinking needs all four. Screens-only systems work, but take longer to set up.
A strong answer"Crew scheduling and operations have interfaces, and our model takes a roster file."
6
Decision owner
Is there a named owner of fatigue risk management with a budget to change how it is done?
Why it mattersThe work crosses safety, flight operations, crew planning, the unions and IT. Without one owner it stalls.
A strong answer"Our director of flight operations safety sponsors it, with a budget line."
7
Confidentiality
Do your union fatigue committees know about the idea, and is it clear who may see a reporter's identity?
Why it mattersReports keep coming only while crews trust the system. A rollout the committees support goes faster.
A strong answer"One named gatekeeper; the pilot and flight attendant committees are on the group and have seen the plan."
8
Governance
Do you have a way to approve AI for safety work: a review path, human sign-off and an audit trail?
Why it mattersFatigue science already warns against a single tool deciding: IATA says no fatigue model should be "the sole go/no-go decision-making tool".8 A clear approval path, with people signing, shortens the road to production.
A strong answer"IT security and the safety office have a review path, and every decision needs a named signer."
9
Vision · for reflection
If your FAA inspector asked tomorrow for evidence that last quarter's fatigue reports were acted on, what would you show, and how long would it take?
Why it mattersYour answer lists the gaps. Each gap is a candidate job for software: reading, linking, modelling, drafting or measuring.
A strong answerA short, specific list, for example: "every report linked to its duty, every pattern with an outcome, every change with a before and after."
What your score means
0–6
Explore
Build the baseline first: time a handful of reports worked in depth, count last year's fatigue calls and schedule changes, and name an owner. Talk to the union fatigue committees early.
7–11
Ready to pilot
Run a measured pilot on a past quarter of your own de-identified reports. Compare what the software links and finds with what your team found, and have your people sign every decision.
12–16
Ready to implement
Plan the full desk: integrations, your rules and risk matrix as settings, the gatekeeper and roles, the group's decision packs, measurement and the records the FAA expects.
19 · Questions for any vendor
Eighteen questions to ask before you buy FRMS software or fatigue risk management software
In shortAsk every vendor the same questions, us included. Several kinds of tools already score schedules, take fatigue reports or link them to rosters. The answers show how the work is really done: what is linked, which sleep the model uses, who can see a reporter, and who decides.
Ask
A good answer
A warning sign
Evidence
Does every number on a fatigue case cite its source: the report text, the roster, the actual times, the model run, the rule?
One click from any number to its source
A score or a chart with no trail
Does it link each report to the roster actually flown and the actual out, off, on and in times, or only to the planned pairing?
The days before the report as flown, with actual times
The planned pairing only
The science
Does it run the fatigue model with the sleep the crew member reported, as well as the model's assumed sleep, and explain the difference?
Both runs, side by side, with the difference explained
Assumed sleep only
Which fatigue models does it work with? Can we keep the one we license?
Yours, through a defined interface
Replace your model to use the tool
What validation evidence exists for the model and thresholds on an operation like ours, and does the tool say when a duty is outside the validated range?
Named studies, your thresholds, and a flag on out-of-range duties
"Validated" with no detail
Patterns and fixes
How does it find patterns, by pairing, flight, hotel, station, hour and fleet, and against what baseline rate?
Your own rate per 1,000 duties, with the rule visible and editable
Monthly counts by fleet
Does it check proposed pairing changes against Part 117, §121.467 and our labor contracts, and show illegal options as refused with the rule?
Every option checked; refused ones shown with the rule named
Legality left to the planner after the decision
Can anything change a roster or pairing without a person, or without the group's approval?
No: only a planner publishes, and only an approved change
Automatic schedule changes
How does it measure whether a change worked, and tell falling fatigue from falling reporting?
Before and after for a set period, with fatigue calls watched beside reports
Fewer reports read as success
Confidentiality
Who sees a reporter's identity? Can any role other than the designated gatekeeper reveal it, and is every reveal logged and reported to the crew representatives?
One role, a reason code, a log the committees can see
Analysts and the board see names
How does it keep fatigue reports that are also voluntary safety or SMS reports separate, and share only what the agreement allows?
Kept apart, with only de-identified facts shared
One shared pool of reports
Decisions and records
How are decisions recorded: reason codes, electronic signatures, votes and quorum, minutes?
All four, with names and times
A status change with no signer
Can it produce our plan update and the evidence the FAA asks for, every line traceable, and track the two-year plan clocks and annual training?
Drafted for a person to approve, each line cited, clocks tracked
A spreadsheet export
What does the audit trail record, can we export it, and can its integrity be verified?
Every action by people and software, exportable, tamper-evident
An editable activity log
Your data and AI
Which AI models does it use, where do they run, and is our data used to train anything?
Named models, a region you choose, no training on your data, in writing
"Proprietary AI", no more detail
Fit and proof
Can we test it on a past quarter of our own de-identified reports before we sign?
Yes, with results compared with what our team found
Demo data only
Will you report extraction and linking accuracy by field, not one number?
By field: flight, duty, sleep, rating, factors
One accuracy figure
Will it support all-cargo operations under §121.495, Part 135 operations and our SMS under Part 5, if we have them?
Each rule set as configuration, not a custom project
Part 117 passenger flights only
We found no published independent test of any vendor's accuracy at reading fatigue reports, as of October 2026. A test on your own past reports is the fairest comparison.
20 · How to start
How to start: five steps to a decision you can defend
In shortMeasure first, then test on your own reports, then decide with numbers. None of these steps commits you to a vendor.
Time five reports worked in depth, two simple and three hard. Note the hours spent pulling rosters and times, running the model and finding similar reports. That is your baseline.
Pull last year's numbers: reports, fatigue calls, schedule changes, and how many changes were measured afterwards.
Gather a past quarter of reports, de-identified, with the rosters and times behind them and the approval of your union fatigue committees.
Run a pilot. Compare what the software links and finds with what your team found, and have your people sign every decision.
Decide with three numbers: reports linked correctly by field, patterns found and when, and hours from report to a decision-ready case.
The full working-session questions: 24 questions in eight groups
Bring the fatigue risk manager, flight operations safety, crew planning, the union fatigue committees, training, the SMS owner and IT into one room. Each question says why it matters, what a strong answer looks like, and the warning sign.
A
Scope and volume
A1
How many fatigue reports and fatigue calls do we get a year, by crew group, base and fleet?
Why it mattersSets the size of the job and where patterns are likely.
Strong answer · warning sign
Counts by group, base and month.
One number, nobody sure of the mix.
A2
Which operations are in scope: Part 117 passenger, all-cargo, Part 135, flight attendants?
Why it mattersEach has its own rules: Part 117, the cargo subparts, Part 135 and §121.467.3,9,10
Strong answer · warning sign
A list with the rule set for each.
"Pilots only, for now," with flight attendant reports in the same inbox.
A3
Do we operate, or plan, anything that needs a §117.7 approval?
Why it mattersAn approval needs data specific to the operation.21
Strong answer · warning sign
The route, the limits it exceeds and the data plan.
An approval assumed to be easy.
B
Current process
B1
Take the last report that led to a change: how long from the first report to the decision, and where did the time go?
Why it mattersOne timed case shows the work better than any process map.
Strong answer · warning sign
Days and hours, step by step.
"It depends."
B2
Which roster does our model read today: planned or flown? Which sleep?
Why it mattersPlanned and flown rosters read differently, and assumed sleep may not be real sleep.8,30
Strong answer · warning sign
Flown, with reported sleep where we have it.
Nobody is sure.
B3
How does the fatigue safety action group get its cases, and how are decisions recorded?
Why it mattersThe pack and the record are what the FAA and the committees see.
Strong answer · warning sign
One pack per case, minutes signed.
Slides and email.
C
Results so far
C1
Which schedule changes did we make for fatigue in the last two years, and did each one work?
Why it mattersThe plan must show a way to assess its effectiveness.4
Strong answer · warning sign
A list with before and after.
Changes made, never measured.
C2
Which safety performance indicators do we track, and who reviews them?
Why it mattersCombining several measures gives the most useful picture.32
Strong answer · warning sign
Reports per 1,000 duties, fatigue calls, time to decision.
Report counts only.
C3
If reports fell 30% next quarter, how would we know whether fatigue fell or reporting did?
Why it mattersA falling report count can hide a falling trust.
Strong answer · warning sign
Fatigue calls and surveys checked alongside.
"Fewer reports is good news."
D
Plans, training and the FAA
D1
When are our pilot and flight attendant plan updates due?
Why it mattersEach must be updated every two years.4,5
Strong answer · warning sign
Two dates, and a draft in progress.
Nobody owns the date.
D2
How do we track annual fatigue training for every group §117.9 names?
Why it mattersPilots, dispatchers, schedulers, operational control staff and their managers all need it.3
Strong answer · warning sign
Completion by group, with reminders.
A spreadsheet updated once a year.
D3
What did our principal operations inspector last ask for on fatigue, and how long did it take to answer?
Why it mattersShows what evidence is expected and how ready it is.
Strong answer · warning sign
A pack, built in a day.
Weeks of digging.
E
Confidentiality and the unions
E1
Who can see a reporter's name today, and is every look recorded?
Why it mattersOne exposed reporter can stop the reports.
Strong answer · warning sign
One role, logged.
Anyone with access to the reporting tool.
E2
How do fatigue reports and voluntary safety reports interact under our agreement?
Why it mattersThe agreement sets what may be shared, and with whom.28
Strong answer · warning sign
A written rule for cross-filed reports.
Handled case by case.
E3
Are pilot and flight attendant representatives on the group, and do they vote?
Why it mattersPilot associations ask for crew representatives on the group.29
Strong answer · warning sign
Both, voting.
Invited, not voting.
F
Systems and data
F1
Where do rosters as flown and actual times live, and can software reach them?
Why it mattersSets the integration work and the timeline.
Strong answer · warning sign
Interfaces we own.
Reports printed from screens.
F2
How do we run our fatigue model today, and can it take a roster from another system?
Why it mattersKeeping your model avoids a second validation.
Strong answer · warning sign
A roster file or an interface.
Re-keyed by hand.
F3
Where are layover hotels and rooms recorded?
Why it mattersMany patterns are hotels, not pairings.
Strong answer · warning sign
Bookings with hotel and room data.
Not recorded.
G
Governance and decision
G1
Which approvals does AI in safety work need here: IT security, the safety office, the unions?
Why it mattersA known path shortens the road to production.
Strong answer · warning sign
A known path and owners.
First time anyone has asked.
G2
Who sponsors the change, and what budget can move?
Why it mattersA pilot without a sponsor ends as a demo.
Strong answer · warning sign
A named executive and a budget line.
No sponsor.
G3
What would a pilot have to show for us to call it a success?
Why it mattersA measured pilot settles the decision.
Strong answer · warning sign
Linking accuracy by field, patterns found, hours to a case.
"Let's see what it does."
H
Vision
H1
What if every fatigue report were linked to the night actually flown on the day it arrived?
Why it mattersTests the appetite for a same-day standard.
Strong answer · warning sign
"Our analysts would judge, not hunt."
"We only need the serious ones."
H2
What if every change came with a measured result 90 days later?
Why it mattersLinks the program to evidence the FAA and the crews can see.
Strong answer · warning sign
The committees would back it.
Nobody owns the follow-up.
H3
Which steps would we never let software do without a person?
Why it mattersTells you where the signatures belong.
Strong answer · warning sign
Revealing a reporter, deciding a case, approving and publishing a change.
A data-driven way for an airline to find and fix the causes of crew fatigue and to show the fixes work. In the US it is defined in 14 CFR 117.3 and is the route to exceed a Part 117 limit, with FAA approval and "at least an equivalent level of safety" (14 CFR 117.7). ICAO (Annex 6 and Doc 9966) and EASA (ORO.FTL.120) use the same idea.
What is the difference between an FRMS and an FRMP?
Every Part 121 airline must keep an FAA-accepted fatigue risk management plan (FRMP) for its pilots, under Public Law 111-216 §212, and for its flight attendants, under the FAA Reauthorization Act of 2018 §335, with annual fatigue training and an update every 2 years. An FRMS is optional and needed only to operate beyond a Part 117 limit.
What does an FRMS have to include?
Six elements: a fatigue risk management policy, an education and awareness training program, a fatigue reporting system, a system for monitoring flightcrew fatigue, an incident reporting process and a performance evaluation (14 CFR 117.7(b)).
What are the Part 117 rest rules?
At least 10 consecutive hours of rest before a flight duty period or reserve, including 8 uninterrupted hours of sleep opportunity (14 CFR 117.25(e)), and 30 consecutive hours free from duty in any 168 (117.25(b)). If a rest will not give that sleep opportunity, the pilot must tell the airline (117.25(f)).
What is the window of circadian low (WOCL)?
2:00 to 5:59 a.m. during a physiological night, "a period of maximum sleepiness" (14 CFR 117.3). Red-eyes that land in it are a classic fatigue hazard even when every limit is met.
Does Part 117 apply to cargo pilots?
Not unless the airline opts in. Part 117 covers Part 121 passenger operations; all-cargo flights stay under older Part 121 rules (14 CFR 117.1 and 121.470(c)). As of October 2026, bills to change this (H.R. 7191 and H.R. 7526) are in committee. The fatigue risk management plan duty applies to cargo airlines too.
What are the FAA fatigue rules for flight attendants?
Flight attendants are not under Part 117. Since the January 10, 2023 compliance date, a flight attendant scheduled for 14 hours of duty or less must get at least 10 consecutive hours of rest, never reduced (14 CFR 121.467), and each Part 121 airline keeps a flight attendant fatigue risk management plan, updated every 2 years.
How often must a fatigue risk management plan be updated?
Every 2 years, for pilots under Public Law 111-216 §212 and for flight attendants under Public Law 115-254 §335, with each update submitted to the FAA for review and acceptance. FAA guidance expresses the cycle as 24 calendar months. Fatigue training under 14 CFR 117.9 is annual.
Are pilot fatigue reports confidential?
They should be. A fatigue reporting system is a required FRMS element, and the FAA's SMS rule requires a confidential employee reporting system "without concern of reprisal" (14 CFR 5.71(a)(7)). Many airlines also take fatigue reports through ASAP, the Aviation Safety Action Program (AC 120-66C).
What is a fatigue call?
In airline usage, a crew member removing themselves from a trip because they are too fatigued to fly safely, usually covered by a reserve. Part 117 bars a pilot from continuing once they report too fatigued (14 CFR 117.5). How fatigue calls are paid and reviewed is set by each airline and its labor contracts, not by an FAA rule.
What is a biomathematical fatigue model, and can it decide on its own?
Software that estimates alertness over a schedule from sleep and body-clock science. IATA's April 2025 guidance says "a model is not a Fatigue Risk Management System", should not be "the sole go/no-go decision-making tool", that outputs "represent the population average", and that model sleep predictions "may not reflect actual sleep".
What is a fatigue safety action group?
The cross-functional group of management, pilot and flight attendant representatives, scheduling and safety staff that reviews fatigue data and decides on mitigations. FAA guidance (AC 120-103A) and ICAO describe it, and a Commission regulation adopted September 8, 2026, not confirmed in force as of October 2026, would name it in EU law.
What changed with safety management systems in 2024?
The FAA's April 26, 2024 rule extended safety management systems (14 CFR Part 5) to Part 135 and air tour operators, who must comply by May 28, 2027. Part 121 airlines had to revise their SMS by May 28, 2025.
Can AI help with fatigue risk management?
It can do the reading and linking: extract each report, attach the roster flown and the actual times, run the model, find patterns and draft the case. It should not reveal reporters, decide schedule changes or replace the fatigue safety action group's judgment; IATA warns that no fatigue model should be "the sole go/no-go decision-making tool".
What should FRMS software do?
Link every fatigue report to the roster actually flown and the actual times; run your fatigue model with reported as well as assumed sleep; find patterns against your own rate; test changes against Part 117, the flight attendant rest rule and your contracts; seal reporters' identity; record decisions with signatures; measure every change; and produce plan updates and FAA evidence. Test it on your own past reports before you buy.
22 · One-page summary
Airline fatigue risk management: one-page summary
Print this page
The readiness questions
About how many fatigue reports do we receive a year?
How many are linked to the roster flown and a model run?
Do we know hours per report, report rate and days to decision?
Is there a date that matters: a plan update, an FAA question, an approval, 2027 SMS?
Can software reach rosters flown, actual times, reports and our model?
Is there a named owner with a budget?
Do the union fatigue committees support it, with one gatekeeper for identity?
Do we have a way to approve AI, with sign-off and an audit trail?
If the FAA asked tomorrow for evidence that last quarter's reports were acted on, what would we show?
Score 0–2 on questions 1–8: 0–6 explore · 7–11 pilot · 12–16 implement.
Key facts as of October 2026: Part 117 in force since January 4, 2014 · pilot and flight attendant plans updated every 2 years · fatigue training annual · Part 135 SMS by May 28, 2027 · all-cargo flights outside Part 117 unless they opt in.
12 terms
FRMS
Fatigue risk management system: data-driven, and in the US the route Part 117 provides (§117.7) beyond its limits.
FRMP
Fatigue risk management plan: required of every Part 121 airline; updated every 2 years.
Part 117
The FAA's flight, duty and rest rules for pilots on passenger flights.
FDP
Flight duty period: from report until the aircraft is parked after the last flight.
Sleep opportunity
Time when sleep is possible; the rules protect it, not the sleep.
WOCL
Window of circadian low: 2:00 to 5:59 a.m. body-clock time.
Acclimated
Adjusted to a time zone: 72 hours there, or 36 hours off duty.
Samn-Perelli
A 7-point fatigue scale, from 1 "fully alert" to 7 "completely exhausted".
Fatigue call
Removing yourself from a trip as too fatigued; usually covered by a reserve.
FSAG
Fatigue safety action group: reviews fatigue data and decides on changes.
SMS · Part 5
Safety management system; fatigue is one hazard within it.
Accessed October 2026. Regulatory status statements are as of October 2026. Where a primary document could not be fetched, the trade or law-firm source that reported it is listed.
NTSB findings, Colgan Air flight 3407, adopted February 2010, and AIN coverage, February 23, 2010. faa.gov; ainonline.com
NTSB, UPS flight 1354 investigation (DCA13MA133), and the 2019-2020 Most Wanted List fatigue sheet. ntsb.gov; Most Wanted List sheet
eCFR, 14 CFR Part 117, Flight and Duty Limitations and Rest Requirements: Flightcrew Members (current text). ecfr.gov
Public Law 111-216, Airline Safety and Federal Aviation Administration Extension Act of 2010, §212. govinfo.gov
Public Law 115-254, FAA Reauthorization Act of 2018, §335. govinfo.gov
Final Supplemental Regulatory Impact Analysis for all-cargo operations, 79 FR 72970, December 9, 2014. federalregister.gov
H.R. 7191, Fatigued Pilot Protection Act, and H.R. 7526, Safe Skies Act of 2026 (119th Congress); ALPA legislative update, August 2026. congress.gov; govinfo.gov; alpa.org
IATA Human Factors Task Force, "Uses and Limitations of Biomathematical Fatigue Models", April 2025. iata.org
14 CFR 121.470 and 14 CFR 121.495 (all-cargo flight time limitations and their fatigue risk management system). ecfr.gov; law.cornell.edu
eCFR, 14 CFR 121.467, Flight attendant duty period limitations and rest requirements. ecfr.gov
IFALPA, position paper 21POS01, Fatigue Risk Management Systems. ifalpa.org
Swedish Transport Agency, Compliance Checklist 2, Subpart FTL for FRM, version 2.0, September 18, 2018 (reproducing ORO.FTL.120). transportstyrelsen.se
ICAO, Fatigue management resources: Doc 9966 (Second Edition, Version 2, Revised, 2020) and the fatigue management guides. icao.int
NTSB, news release retiring the Most Wanted List, December 14, 2023. ntsb.gov
"Flightcrew Member Duty and Rest Requirements", final rule, 77 FR 330, January 4, 2012. federalregister.gov
Technical correction, 78 FR 11090, February 15, 2013: effective date January 4, 2014. federalregister.gov
"Flight Attendant Duty Period Limitations and Rest Requirements", final rule, 87 FR 61452, October 12, 2022 (effective November 14, 2022; compliance January 10, 2023). federalregister.gov
FAA, proposed rule "Ensuring Passenger Safety by Preempting Duty and Rest Requirements", July 6, 2026. federalregister.gov
FAA Advisory Circular 120-103A, Fatigue Risk Management Systems for Aviation Safety, May 6, 2013. faa.gov
FAA Advisory Circulars 117-2 (fatigue education and awareness training) and 117-3 (fitness for duty), October 11, 2012. AC 117-2; AC 117-3
FAA information collection 2120-0751, "Flight and Duty Limitations and Rest Requirements: Flightcrew Members", April 30 and July 1, 2026. federalregister.gov; federalregister.gov
FAA information collection 2120-0789, "Flight Attendant Fatigue Risk Management Plan", June 18 and August 20, 2026. federalregister.gov; federalregister.gov
FAA Information for Operators: InFO 10013 and 10017 (fatigue risk management plans), InFO 13008 (required updates to FRMP), InFO 15001 (fatigue education and awareness training), InFO 19007 (flight attendant FRMP). InFO 13008; InFO 15001; InFO 19007
IFALPA, FRMS checklist for member associations; ALPA, fatigue safety action group presentation. ifalpa.org; alpa
Devine J.K., Choynowski J., Hursh S.R., "Evaluation of a Biomathematical Modeling Software Tool for the Prediction of Risk in Flight Schedules Compared Against Incidence of Fatigue Reports", Safety 11(1):4, January 7, 2025. doi.org
ICAO, news release on the adoption of FRMS Standards, 2011, and Annex 6 Part I fatigue management excerpt. icao.int; Annex 6 excerpt
IATA, "Fatigue Safety Performance Indicators (SPIs)", 2014; ICAO, IATA and IFALPA, Fatigue Management Guide for Airline Operators, second edition, 2015. icao.int; iata.org
Council of the European Union, document 12992/26, September 10, 2026: Commission Delegated Regulation C(2026) 6142 of September 8, 2026. consilium.europa.eu; annex
Next step
See it on your own fatigue reports.
Explore Botminds Fatigue Risk Management System, or talk to us about a measured pilot on a past quarter of your own de-identified fatigue reports.
We use essential cookies to make this site work, and cookieless, anonymised analytics to understand how it's used. "Accept all" also turns on Google Analytics, which sets cookies. Privacy policy