The Big Picture

In the early 2000s, the experimental amateur-built fleet averaged about 4.33 fatal accidents per 100,000 flight hours — more than four times the certified general aviation rate. By 2024, that number had fallen to 2.04 per 100,000 hours. That is a 53% decline in the fatal accident rate over roughly two decades.

This is not a statistical fluke. The improvement shows up decade over decade, across multiple metrics. Fewer engine failures, fewer builder errors, fewer fatal outcomes when accidents do occur. The community is building better airplanes, installing better engines, and the older, less reliable designs are gradually aging out of the fleet.

Fatal Rate by Decade

Fatal accidents per 100,000 flight hours, compared to certified general aviation (GA):

Sources: NTSB accident data, FAA GA activity surveys for flight hour estimates. Certified GA rate uses the NTSB benchmark of approximately 1.0 fatal accidents per 100K hours.

The Convergence Story

The gap between E-AB and certified GA has narrowed from 4.3x in the 2000s to 2.0x in 2024. At the current trajectory, experimental aircraft are approaching certified GA safety levels faster than most people realize.

The improvement is coming from three reinforcing trends: better engines (especially the shift away from two-strokes to the Rotax 912), better kit designs with clearer instructions and quick-build options, and the natural retirement of older, less reliable designs from the active fleet. Each decade, the fleet gets younger and more standardized.

Will E-AB ever match certified GA? Maybe not completely — the diversity of the fleet and the builder-maintained model introduce inherent variability. But the direction is unmistakable, and the pace of improvement is encouraging.

What Is Driving the Improvement

The decline is not uniform across all accident causes. Some categories have improved dramatically; others have barely budged.

The Engine Revolution

The single biggest driver of improvement is the shift away from unreliable two-stroke engines to the Rotax 912 family. In the early 2000s, two-stroke-powered ultralights and light experimental aircraft dominated the accident statistics. Two-strokes are mechanically simpler but inherently less reliable in sustained aviation use — they run hotter, have shorter TBOs, and are more sensitive to fuel quality and carburetor icing.

The Rotax 912 has the lowest engine-initiated failure rate in our dataset. As the fleet has shifted toward Rotax-powered designs (and to a lesser extent, toward well-proven Lycoming and Continental installations in kit aircraft), the mechanical-cause accident rate has dropped substantially. The engine you choose is no longer the gamble it once was — if you pick a proven powerplant.

Better Kits, Fewer Builder Mistakes

Builder-error accidents have declined as the kit industry has matured. The kits of the 1990s often shipped with minimal instructions and required significant builder ingenuity. Modern kits from manufacturers like Van’s Aircraft, Zenith, Sonex, and CubCrafters come with detailed step-by-step instructions, factory support lines, and quick-build options that reduce the scope for construction errors.

The rise of builder assist centers and the EAA’s technical counselor program have also helped. More builders now have their work inspected during construction, not just at the final airworthiness inspection. The data reflects this: fewer fuel system errors, fewer control rigging mistakes, and fewer first-flight surprises.

Accidents Are Becoming More Survivable

Even when accidents happen, a smaller percentage are fatal:

The fatal percentage has declined from 26.7% in the 2000s to 22.0% in the 2020s. This likely reflects better crash structures in modern kit designs, wider adoption of shoulder harnesses and energy-absorbing seats, improved ELT technology for faster rescue response, and the growing market share of slower, more forgiving STOL designs like the Zenith CH 750 and CubCrafters Carbon Cub.

When a slow STOL aircraft has an accident, the pilot walks away more than 90% of the time. As these designs gain market share, the fleet-wide survivability rate improves even if the raw accident count stays flat.

What Has Not Improved

The hardware has gotten better. The pilot decision-making? Not as much. Some accident categories remain stubbornly persistent:

Stall/spin remains the number one killer in the E-AB fleet, and the rate has not dropped as sharply as engine-related causes. VFR into IMC continues to be nearly always fatal — better avionics and weather information have not translated into better pilot decisions when it counts. And low-altitude maneuvering and showing off at airshows still kills at roughly three-quarters fatal. You cannot engineer out bad judgment with a better kit or a better engine.

Do Not Overclaim

The trend is real and encouraging, but several caveats apply:

• 2025 data is partial (only through approximately March), so any rate computed for the current year is unreliable and should not be cited as a trend point.
• Year-to-year variation is significant. The rate spiked to 4.78 per 100K hours in 2014 before dropping again. Multi-year averages are more meaningful than any single year.
• The improvement is mostly in engine and mechanical categories. Pilot-error rates are more stubborn. If engine reliability plateaus (and it may, since the Rotax 912 is already excellent), future improvement will require tackling pilot decision-making — a harder problem.
• Certified GA has also been improving. The certified fleet’s rate dropped from about 1.2 to about 0.9 over the same period. So the gap may not close as fast as the raw numbers suggest — E-AB is chasing a moving target.
• Flight hour estimates are approximations. The FAA does not precisely track experimental aircraft hours. The per-100K-hour rates rely on estimates from the GA Activity Survey, which introduces uncertainty.

Year-by-Year Accident Counts

Total and fatal E-AB accidents by year from our NTSB dataset. Note the general downward trend in fatal counts despite a growing fleet:

2025 is excluded because data collection is ongoing and the year is incomplete.

What This Means if You Are Building or Buying

The improving safety trend is good news, but it is not automatic. The builders and pilots benefiting from these gains are the ones who:

• Choose proven engines. The Rotax 912, Lycoming O-320, and Continental O-200 have decades of service data. The improvement in the fleet-wide rate is largely because more builders are choosing these over auto conversions and two-strokes. See our engine reliability comparison.
• Build from well-supported kits. A modern kit with good instructions, factory support, and a large builder community is a fundamentally different proposition than a plans-built project from a defunct designer. Explore safety data for all aircraft types.
• Fly conservatively. The hardware improvements have done their part. The remaining risk is overwhelmingly in pilot decision-making — stall/spin awareness, weather discipline, and resisting the urge to show off at low altitude. Estimate your personal risk profile.
• Pick the right aircraft for the mission. The safest experimental aircraft are slow, forgiving STOL designs. The most dangerous are fast composites. If your mission does not require 200 knots, you can avoid the most dangerous part of the fleet entirely.

The Bottom Line

Experimental aircraft are meaningfully safer than they were 25 years ago. The fatal accident rate has roughly halved. The gap with certified GA has narrowed from 4.3x to 2.0x. Better engines, better kits, and fleet turnover are all contributing.

But the remaining risk is concentrated in pilot decisions, not hardware failures. Stall/spin, VFR into IMC, and low-altitude maneuvering kill at the same rates they always have. Closing the rest of the gap will require changes in how we fly, not just changes in what we fly.

The trend is encouraging. The work is not done. Both of those things can be true at the same time.

Methodology

• Accident data: 8,817 NTSB E-AB accident records, 1982-2026. Each accident is classified by initiator (root cause) using our automated classification system.
• Per-100K-hour rates use FAA General Aviation Activity Survey estimates for E-AB flight hours, cross-referenced with NTSB fatal accident counts.
• Certified GA comparison uses the NTSB-published benchmark of approximately 1.0 fatal accidents per 100,000 flight hours, which has itself declined over the period studied.
• Decade averages smooth year-to-year variation. Individual years can deviate significantly from the trend.
• This analysis covers the E-AB (Experimental Amateur-Built) category only. Other experimental categories (E-AB racing, exhibition, etc.) are excluded.