A probabilistic neural network was developed to classify corrosion damage depth ranges based on the aluminum alloy failure mode. The results obtained indicate that corrosion damage depth for aluminum alloy can be classified into three groups. Statistical study on classified corrosion damage was carried out. The results show that pitting corrosion depth for aluminum alloy is in conforms to Gumbel distribution. The normal distribution fits well with intergranular corrosion depth and the exfoliation corrosion depth is consistent with Weibull distribution law. It may be necessary to use several distribution functions rather than a single distribution to represent corrosion damage characteristics due to the large distribution of corrosion depth in aircraft materials. According to corrosion damage depth distribution, corrosion depth was simulated by Monte Carlo method and used as the starting crack size. Fatigue lives were estimated by using a life prediction program AFGROW and the results are in good agreement with the experimental data. A probabilistic analysis shows that the distribution of fatigue lives is strongly correlated to the distribution of corrosion damage depth and should be classified into several groups to study.