We examined the utility of two conceptual taxonomies of human error in describing and explaining the erroneous actions of pilots involved in aviation accidents. A comprehensive review of all U.S. Navy/Marine Corps aviation accidents between January 1977 and 1992 was conducted. This review revealed 1,518 mishaps that implicated an action or condition of the pilot to be a definite causal factor. The specific type of pilot-causal factors identified for each mishap was then placed in a traditional information processing model and the model of unsafe acts proposed by Reason (1990, Human Error, Cambridge University Press). Of the pilot-causal factors that fit within an information processing model (82%), over two-thirds (73.22%) fell within the decision and response execution stages of the model. The response execution stage alone accounted for over 45% of the pilot-causal factors that fit into this model. Of the pilot-causal factors that fit within the model of unsafe acts (86%), over 58% were classified as intended action mistakes (knowledge- and rule-based mistakes). These preliminary findings suggest that several aviation accidents may be due to pilots' misapplication of cognitive rules to solve problems encountered prior to the mishap. Training that allows actions to become skilled-based rather than rule- or knowledge-based may be needed.