Hours Aren't Armor
Once an accident starts, logged hours hardly change the outcome.
Among 9,028 pilots in personal-flight accidents, the 1,454 fatal accidents had pilots with a median of 900 hours. Non-fatal accidents had a median of 865.
Pilot hours roughly follow a lognormal curve
Hours span three orders of magnitude, so the axis is logarithmic. A bell curve on this axis is a lognormal distribution.
f(h) = 1 / (h·σ·√(2π)) · exp(−(ln h − μ)² / 2σ²)
μ = 6.851, σ = 1.518, fitted median ≈ 945 h
Fatal and non-fatal accident pilots look almost the same
If experience strongly protected pilots in an accident, these two curves would pull apart.
The chance an accident is fatal, by the pilot's hours
The shaded band is a 95% bootstrap range. Daylight and good weather lower the whole curve, and hours barely bend it.
Does this mean experience doesn't make you safer?
That was my first thought too: if most accidents come down to pilot error and hours don't change the outcome, maybe pilots don't make fewer mistakes as they gain experience. The data doesn't say that. It only includes flights that ended in an accident, so it can't show how often pilots make mistakes, or how many accidents experience prevents.
Experience does seem to change the mistakes behind accidents. Investigators cite aircraft control in 49% of accidents for pilots under 100 hours, and in 38% at 10,000+. Decision-making shows no clear trend, staying around 9–13% at every level.
| PIC hours | Aircraft control | Decision-making |
|---|---|---|
| Under 100 | 49.1% | 11.0% |
| 100–249 | 46.0% | 12.2% |
| 250–499 | 42.1% | 12.6% |
| 500–999 | 42.1% | 11.8% |
| 1,000–1,999 | 39.8% | 8.5% |
| 2,000–4,999 | 38.6% | 11.6% |
| 5,000–9,999 | 31.4% | 10.8% |
| 10,000+ | 38.0% | 10.3% |
One way to read it: experience seems to change the mistakes you make more than it changes what happens once one turns into an accident.
See the work
The analysis, with its data frozen so the results don't change, is available as a notebook. AI was used extensively to generate the notebook, guided by some of my knowledge acquired in from a data analytics class.
US Part 91 personal flights in certificated piston singles, 2008–2024, one-aircraft accidents with the pilot's hours on file. NTSB accident data. These figures describe pilots already in an accident. They don't show whether experienced pilots have fewer accidents. Causes come from accidents through 2023 with a cause finding, and they're investigators' judgments rather than measurements.