Lyons TJ, Nace W. Aircraft crash rates and cumulative hours: USAF data for 25 airframes, 1950-2006. Aviat Space Environ Med 2007; 78:923–5. Introduction: Long-term trends in aircraft crash rates are of interest to researchers in aerospace medicine, the acquisition community, the aircraft industry, and the flying public. Methods: Data were obtained from the USAF Safety Center database for all major accidents (Class A Mishaps) and for annual flight hours for fiscal years 1950 through 2006. Regression analysis was accomplished on the logarithmic transformation of the cumulative flight hours and cumulative crash rate for each aircraft. Results: The relationship between crash rate, y, and flying hours, t, is approximately linear on logarithmic coordinates, so it can be expressed as a power relation, y = bt^m, where the exponent, m, reflects the slope. The model fit was significant for all 15 USAF fighter, attack, and training aircraft studied (r² = 0.78–0.99) with all aircraft showing a negative exponent ranging from −0.25 to −0.75. This power function with a negative exponent model also fit four of five bomber aircraft and four of five transport aircraft studied. The Predator unmanned aircraft also fit this model with a slope of −0.37. Discussion: This study exposes a strong inverse relationship between cumulative flight hours and crash rate for most USAF aircraft, indicating a high degree of organizational improvement. Learning was not confined only to aircrew, as evidenced by similar improvements in unmanned aircraft. The data do not support an increasing risk of crashes as aircraft age. Further research should be conducted on different flying organizations to further define the characteristics of organizational learning in aviation.