The fatigue crack growth tests for 2024-T62 aluminum alloy were conducted under three stress levels, the dispersion of fatigue crack growth were analyzed by using experimental data based on mathematical statistics. The analysis results indicate that the experimental data of fatigue crack growth obey the log-norm distribution. The standard deviation and coefficient of variation（COV） both increase with the loading cycle raising. The stress levels have some effect on dispersion of fatigue crack growth sizes, when the loading cycle increases, the standard deviation of fatigue crack growth under high stress level is greater than the standard deviation of fatigue crack growth under low stress level, and the increasing speed of standard deviation and COV of fatigue crack growth under high stress level is faster than the increasing speed of standard deviation and COV of fatigue crack growth under low stress level. The stochastic fatigue crack growth model is verified by experimental data.