A House of Cards at 30,000 Feet: What 2 Million Safety Records Reveal About the People Keeping You Alive

The story isn't that flying is dangerous.

Per flight hour, U.S. aviation was the safest it had ever been. The general aviation fatal accident rate hit a historic low in 2023 — 0.65 per 100,000 flight hours, less than half the rate of the 1990s. No Part 121 airline passenger died in a crash between February 2009 and January 2025 — a 16-year streak without precedent in the history of commercial aviation. Per passenger mile, flying is roughly 1,200 times safer than driving. Fatalities across all of aviation dropped 60 percent in 25 years.

That is not the story. The story is that every layer of the system designed to keep it safe is simultaneously breaking down.

AviatorDB analyzed 1,971,811 federal aviation safety records — the FAA's mandatory maintenance filings, the NTSB's accident database, and NASA's confidential safety reporting system — and found a pattern that no single dataset reveals on its own. The controllers who separate aircraft are stretched thinner than at any point in modern history. The mechanics who keep aging fleets airworthy are retiring faster than they can be replaced. The manufacturer that builds most of America's commercial aircraft cannot keep up with demand. And the regulator responsible for holding it all together has, by its own inspector general's assessment, failed to do so.

The common thread is not technology. It is not regulation. It is people. There are not enough qualified people left to do the work — and the data shows exactly what happens when they leave.

Near-mid-air collisions have tripled in a decade. That is a symptom. This is the disease.

The People Are the System

Aviation safety is not a product of technology alone. It is a product of people — the controllers who separate aircraft, the mechanics who inspect them, the pilots who fly them, and the inspectors who oversee the operation. Technology helps. Automation helps. But when the engine of a Boeing 737 sheds a fan blade at 32,000 feet, it is a mechanic who finds the fatigue crack before the next flight. When two aircraft converge on the same runway, it is a controller who sees it and acts. When a door plug works itself loose over 154 flights, it is an inspector who is supposed to catch it.

The data in 1.97 million federal safety records tells one story above all others: the experienced workforce that built American aviation's safety record is leaving, and the system is not replacing them fast enough.

The Controllers: Sixteen Percent Fewer, Watching Ten Percent More

In 2011, the FAA employed 15,627 certified air traffic controllers. By the end of fiscal year 2025, that number had fallen to 13,164 — a 16 percent decline. Over the same period, U.S. air traffic grew approximately 10 percent, from roughly 16 million annual flights to nearly 17 million.

The ratio of controllers to flights dropped from 977 per million flights in 2011 to 774 in 2025 — a 21 percent decline. Each controller is now responsible for more aircraft in more complex airspace than at any point in modern history.

A Government Accountability Office report published in December 2025 confirmed that 285 of 313 U.S. ATC facilities — 91 percent — operate below recommended staffing levels. The 19 largest facilities, which handle 27 percent of commercial operations and 40 percent of delays, are staffed 15 percent or more below target. A NASA study found that 76 percent of controllers work schedules that lead to chronic fatigue, averaging 5.8 hours of sleep per workday. Six-day mandatory workweeks are common. The NTSB linked the Austin-Bergstrom near-miss of February 2023 to a controller working overtime during one such six-day stretch.

The hiring pipeline cannot keep up. More than 200,000 people have applied to become controllers, but only approximately 2 percent complete both the hiring process and the training, which takes two to six years with substantial attrition. In February 2025, the FAA laid off approximately 400 employees in agency-wide cuts. No controllers were directly fired, but safety-supporting roles were eliminated: aviation safety assistants, maintenance mechanics, and roughly 12 percent of aeronautical information specialists.

In June 2025, the National Academies of Sciences released a congressionally commissioned report on the controller workforce. Among its recommendations: that the FAA conduct research to determine the relationship between facility staffing levels and safety outcomes.

The Mechanics: A Retirement Cliff With No Safety Net

The people who keep aircraft airworthy are disappearing even faster.

Twenty-seven percent of FAA-certificated airframe and powerplant mechanics are over the age of 64. Industry analysts project that 80 percent of the current maintenance workforce will retire within five to six years. The FAA issued 9,013 new mechanic certificates in 2024 — but the industry needs at least 13,100 per year to maintain pace, creating an annual shortfall of roughly 4,000. The projected deficit reaches 40,000 unfilled positions by 2028, representing approximately a 20 percent workforce shortage.

This is not a problem that appeared overnight. It is the result of decades of decisions. Heavy maintenance outsourced by the nine major U.S. carriers grew from 34 percent in 2003 to 71 percent by 2007. Foreign repair stations increased from 21 to 27 percent of outsourced work. Total maintenance jobs at U.S. passenger carriers fell from 72,211 in 2000 to 50,580 in 2011 — a 30 percent decline.

The work moved offshore. The jobs disappeared. The pipeline of young Americans entering the trade dried up. And now the generation that stayed is retiring.

The Congressional Research Service warned that "increased outsourcing and offshoring may contribute to a dangerous nexus between cost-cutting and weaker regulatory oversight." The FAA's own inspectors found foreign facilities using outdated maintenance manuals, failing tool calibration requirements, and mixing scrapped parts with usable inventory. In December 2024, the Department of Transportation launched an investigation into FAA oversight of foreign maintenance contractors after an audit found a Hawaiian Airlines maintenance facility in the Philippines did not meet U.S. regulatory standards. Frontier Airlines mechanics reported finding passenger seats installed in wrong slots and hydraulic parts installed backwards on aircraft returning from outsourced heavy maintenance.

When 80 percent of experienced mechanics retire in the next five years and the replacement pipeline produces only 69 percent of what is needed, who maintains the fleet?

The Fleet: Aging Aircraft That Cannot Be Replaced

That question becomes more urgent when you consider what those mechanics are working on.

The average age of the U.S. commercial fleet has risen to approximately 15 years, up from roughly 11 years in the 1990s. United Airlines operates more than 200 aircraft over 25 years old, with a median fleet age of 19.4 years. Delta's median is 17.2 years.

Airlines want newer aircraft. They cannot get them. Boeing, the sole domestic large-aircraft manufacturer, has a backlog of approximately 17,000 orders — a wait of 10 to 11 years. After the Alaska Airlines 737 MAX 9 door plug blowout in January 2024, the FAA capped 737 MAX production at 38 per month. Airlines received roughly half their contractual deliveries.

The NTSB's final report on the door plug incident found that four bolts were never installed at Boeing's Renton, Washington factory — by workers who were not the required specialized technicians, with no documentation created and no quality assurance inspection performed. An FAA audit of Spirit AeroSystems, Boeing's primary fuselage supplier, found the company failed seven of 13 product audits — dented fuselages, missing fasteners, foreign object debris, a wrench left inside a component. In April 2024, Boeing engineer Sam Salehpour told the Senate that 98.7 percent of inspected 787s had gaps exceeding specifications. He described workers "jumping on pieces of the airplane" to force alignment on the 777 production line.

Boeing is a staffing story too. Experienced production workers left during the pandemic. Quality control expertise walked out the door. The company pled guilty to conspiracy to defraud the United States and paid $1.1 billion in fines and settlements related to the 737 MAX crashes.

So airlines are keeping older aircraft flying longer. AviatorDB's analysis of 1.73 million FAA Service Difficulty Reports (SDRs) shows what that means: corrosion and cracking now account for 41.2 percent of all maintenance filings — 706,076 reports. Door-related SDRs increased 39.8 percent comparing 2020-2024 to 2015-2019. Window SDRs increased 45.6 percent. Parts separating from aircraft in flight doubled from 19-33 per year in 2016-2017 to 50-60 per year by 2022-2024.

These structural trends are growing against flat total SDR volume of 60,000 to 67,000 reports per year, meaning structural failures are consuming a larger and larger share of the maintenance burden — on aircraft that are getting older, maintained by a workforce that is getting smaller, with replacement aircraft that are not coming.

What Happens When the People Leave: The Data

The consequences of this workforce erosion are already visible in the data. They show up as near-misses.

Near mid-air collisions reported to NASA's Aviation Safety Reporting System have tripled in a decade. In 2015, pilots filed 308 NMAC reports, representing 5.2 percent of all ASRS filings. By 2025, that number reached 660 — 12.0 percent of all reports.

This is not a function of more flights. U.S. air traffic has been roughly flat at 16 to 17 million annual flights since 2015. The rate of near-misses per flight is rising.

One hundred and one ASRS reports during this period described aircraft with both zero horizontal and zero vertical separation — aircraft that effectively occupied the same point in space.

The 2020 data makes the trend harder to dismiss. When COVID-19 reduced air traffic by approximately 63 percent, near-mid-air collision reports dropped by only 8 percent. Fewer planes were flying, but planes were still nearly hitting each other at almost the same rate. The per-flight NMAC rate actually increased — from 6.4 percent of reports in 2019 to 7.6 percent in 2020. Removing aircraft from the sky did not fix the problem. Something structural had changed in the system's ability to keep them separated.

AviatorDB's cross-referencing of controller staffing data against near-miss reporting trends may represent the first public analysis connecting these two datasets. No federal agency has formally studied this relationship. The National Academies recommended that the FAA do so in June 2025. The data does not prove that fewer controllers caused more near-misses. But when 16 percent fewer people are watching 10 percent more flights, and near-miss reports triple, the correlation demands investigation.

These numbers almost certainly undercount the problem. A 2024 academic study used ADS-B surveillance data at a university airport and identified 16 events in which aircraft came within 500 feet of each other. Not a single one was reported to the FAA or NASA. The researchers estimated that the ASRS captures between 0.9 and 3.3 percent of actual NMAC-level events — meaning the true annual number may be between 12,900 and 27,700.

Drones are adding pressure to a system already stretched thin. In 2024, unmanned aircraft accounted for more than two-thirds of near-mid-air collisions at the 30 busiest U.S. airports. Illegal drone incursions rose 25.6 percent in the first quarter of 2025. During congressional testimony in July 2025, a homeland security official warned: "We are going to have a catastrophic event."

The Airline That Proves the Rule

One carrier in AviatorDB's data offers a counterexample that makes the staffing argument impossible to ignore.

Southwest Airlines files more FAA Service Difficulty Reports than any other carrier — 239,602 total, 59 percent more than second-place American Airlines. On a per-aircraft basis, Southwest files 75 SDRs per airframe annually, the highest in the industry.

And yet Southwest has the lowest serious maintenance event rate of any major U.S. airline: 0.06 percent, compared to an industry average of approximately 0.25 percent. Hawaiian Airlines, by comparison, has a serious event rate nearly nine times higher.

The difference is people. Southwest operates a single-fleet model — exclusively Boeing 737s since its founding — meaning every mechanic specializes in one aircraft type. It maintains the largest in-house maintenance workforce of any U.S. carrier at over 4,600 employees. It was an early adopter of NASA's Aviation Safety Action Program and maintains a formal no-discipline reporting policy. Southwest's 47-year record of one in-flight passenger fatality is unmatched in U.S. aviation.

Southwest reports more because its people catch more. Its people catch more because they know the fleet intimately, work in-house rather than at a contract shop overseas, and are rewarded rather than punished for identifying problems. This is what psychologist James Reason described as "Just Culture" — organizations that encourage reporting without punishment have fewer adverse events.

The rest of the industry moved in the opposite direction. AviatorDB's data shows that the number of operators filing SDRs dropped from 381 in 2011 to 200 in 2025 — a 47 percent decline — while total SDR volume remained flat. Fewer companies, with fewer in-house mechanics, are now responsible for the same volume of maintenance issues. Consolidation concentrates risk. Outsourcing exports expertise.

Southwest kept its people. The data shows what that's worth.

Where the Money Went

None of these workforce crises occurred in a vacuum. They were, in many cases, funded.

During the period from 2013 to 2019 — the airline industry's golden era of profitability — the four largest U.S. carriers earned a combined $80 to $100 billion in after-tax profit. At least $39 billion went to stock buybacks. Not to hiring mechanics. Not to funding ATC modernization. Not to building training pipelines for the next generation of aviation workers.

American Airlines was the most aggressive: $11.9 billion in buybacks against $24.9 billion in operating cash flow, leaving negative free cash flow. By 2023, its stock had fallen 90 percent from peak and debt was crushing the balance sheet. During this same period, the Department of Transportation's Inspector General found that FAA inspectors accepted root cause analyses from American that did not identify the true root cause in 92 percent of sampled cases. In 2024, American's own pilots' union reported a "significant spike in safety- and maintenance-related problems," including tools left in wheel wells and improperly closed maintenance actions.

Then came 2020. The same carriers that had spent $39 billion buying back their own stock sought more than $50 billion in taxpayer bailouts under the CARES Act.

The industry had the money to invest in the workforce that keeps aviation safe. The data shows where they spent it instead.

A House of Cards

Aviation professionals use the Swiss cheese model to explain how accidents happen — not from a single failure, but from weaknesses in multiple defense layers aligning at the same moment. Every complex system has holes. Safety depends on those holes never lining up.

What AviatorDB's analysis reveals is that the holes are not random. They are being created by the same force: the erosion of the experienced human workforce at every level of the system.

• Controllers: 16 percent fewer since 2011, watching 10 percent more flights. 91 percent of facilities understaffed. Near-misses tripled.
• Mechanics: 27 percent over age 64. 80 percent retiring within five years. Pipeline producing 69 percent of replacements needed. 40,000 projected shortfall by 2028.
• Manufacturers: Boeing lost experienced production workers, quality collapsed. 17,000-plane backlog. Airlines cannot replace aging fleet.
• Maintenance: 71 percent of heavy work outsourced offshore. 30 percent fewer in-house jobs. Foreign facilities failing FAA audits.
• Pilots: Structured training pathways producing only one-third of new airline pilots. Regional first officer pay tripled — the market screaming shortage.
• Regulators: FAA accepted flawed root cause analyses from American Airlines 92 percent of the time. Took more than four years to resolve known maintenance violations at SkyWest Airlines. Virtual inspections replacing on-site visits. Chronic understaffing of inspectors.

Every one of these is a people problem. Not enough of them, not enough experience, not enough investment in the next generation.

On January 29, 2025, every layer failed at once. A single controller handled two positions at Reagan National Airport. A military helicopter with an instrument malfunction flew into commercial approach airspace on a route the FAA had left open despite 14 years of documented near-misses. A collision avoidance system issued an alert but could not command evasive action. The crew of a 20-year-old regional jet had 19 seconds of warning. Sixty-seven people died.

The crash was not a surprise. It was the sound of a house of cards being hit by a gust of wind that everyone had seen coming.

What the Data Cannot Tell Us

This analysis has limitations that must be stated clearly.

Correlation is not causation. The relationship between controller staffing levels and near-miss rates has not been formally studied by the FAA, the NTSB, or any federal agency. The National Academies recommended this research in June 2025. Until it is conducted, the parallel trends we identified — fewer controllers, more near-misses — should be treated as a finding that warrants investigation, not a proven causal mechanism.

ASRS reports are voluntary and self-reported. NASA estimates that the system captures a small fraction of actual events. Our NMAC figures represent a floor, not a ceiling.

Service Difficulty Reports document maintenance conditions, not safety of flight. High SDR counts on an airframe reflect diligent reporting, not necessarily an unsafe aircraft. Southwest Airlines' status as the highest filer and the safest operator demonstrates this clearly.

The mechanic workforce projections cited here are industry estimates that carry uncertainty. The actual retirement timeline will depend on economic conditions, pension structures, and individual decisions.

And flying remains, today, extraordinarily safe. The people still in the system are doing remarkable work under increasing pressure. The safety record they have built over decades is real, and it is holding.

But a record built by people can only hold as long as the people are there. The data in 2 million federal safety records shows them leaving — from the control towers, from the maintenance hangars, from the factory floors, from the inspection offices. The system is not failing. It is thinning. And the question is no longer whether the margins will hold. It is how much thinner they can get before they don't.

---

METHODOLOGY

AviatorDB analyzed 1,971,811 records across three federal aviation safety databases:

• FAA Service Difficulty Reports (SDR): 1,727,940 mandatory maintenance difficulty filings from Part 121 and Part 135 operators, spanning 1995 to 2026. Source: FAA SDRx bulk data download.
• NTSB Accident Database: 94,440 accident and incident records spanning 1962 to 2026. Source: NTSB Aviation Accident Database.
• NASA Aviation Safety Reporting System (ASRS): 149,431 voluntary safety reports spanning 2000 to 2025. Source: NASA ASRS database.

Cross-referencing was performed using FAA registration numbers (N-numbers), operator designators, aircraft type codes, date ranges, and geographic identifiers. All queries were run against primary source data; no third-party aggregations were used.

Controller staffing data was sourced from NATCA published workforce statistics and the GAO report GAO-26-107320. Flight volume data was sourced from the FAA and Bureau of Transportation Statistics. Mechanic workforce projections were sourced from Oliver Wyman, ATEC, and Aviation Week industry analyses.

---

SOURCE CITATIONS

1. AOPA McSpadden Report (35th edition, Nov 2025): GA fatal rate 0.65/100K hrs, lowest ever. aopa.org
2. GAO-26-107320: Air Traffic Control Workforce. December 2025. 91% of facilities understaffed. gao.gov
3. National Academies: The Air Traffic Controller Workforce Imperative. June 2025. nationalacademies.org
4. NATCA FY23 Staffing Fact Sheet. natca.org
5. DOT OIG: FAA Controller Staffing at Critical Facilities. June 2023. oig.dot.gov
6. NASA Controller Alertness and Fatigue Monitoring Study. nasa.gov
7. GAO Blog: While Thousands Applied to Become Controllers, There's Still a Shortage. gao.gov
8. Academic study on NMAC underreporting, 2024. sciencedirect.com
9. Congressional testimony on drone incursions, July 2025. homeland.house.gov
10. Drone NMACs at 30 busiest airports. san.com
11. NTSB Final Report AIR2504: Alaska Airlines 737 MAX 9 door plug. June 24, 2025. ntsb.gov
12. Spirit AeroSystems FAA audit failures. npr.org
13. Boeing whistleblower Senate testimony, April 17, 2024. senate.gov
14. Boeing DOJ settlement. cnbc.com
15. FAA production restrictions on 737 MAX. faa.gov
16. Fleet age analysis. thehill.com
17. Oliver Wyman aircraft mechanic workforce analysis. arsa.org
18. ATEC 2024 mechanic certificate data. atec-amt.org
19. Aviation Week: mechanic shortage. aviationweek.com
20. CRS R42876: Airline maintenance outsourcing. congress.gov
21. DOT OIG Report 38631: FAA oversight of American Airlines. oig.dot.gov
22. DOT OIG Report 37731: Southwest airworthiness concerns. oig.dot.gov
23. DOT OIG Report 46882: SkyWest maintenance violations. oig.dot.gov
24. GAO RCED-91-24: Airline SDR usage survey. gao.gov
25. James Reason, "Just Culture" framework (1997). skybrary.aero
26. Senate Commerce Committee hearing on near-misses, November 2023. commerce.senate.gov
27. DOGE/FAA workforce cuts, February 2025. pbs.org
28. Flight Safety Foundation: 132 airliner accidents in 2024, 26% increase. flightsafety.org
29. BTS: US airline financial data. bts.gov
30. Nasdaq: Airline buybacks analysis. nasdaq.com
31. AFA-CWA: No Stock Buybacks in Aviation. afacwa.org
32. FAA Air Traffic By The Numbers FY2024. faa.gov
33. BTS: Pilot-reported NMACs by degree of hazard. bts.gov
34. FAA SDR Reporting System. sdrs.faa.gov
35. AviatorDB internal analysis: 1,971,811 records across faa_sdr (1,727,940), ntsb_accidents (94,440), nasa_asrs (149,431).

---

AviatorDB (aviatordb.com) is an aviation data platform used by plane spotters, researchers, journalists, and aviation enthusiasts worldwide. The platform aggregates and cross-references U.S. and international aviation databases — including FAA registration, NTSB accident records, NASA safety reports, and maintenance filings — covering more than 767,000 aircraft across 200 countries and tracking over 250,000 aircraft positions daily. This analysis was conducted independently and is not affiliated with any government agency, airline, or aircraft manufacturer.