Training programs are increasingly relying on high level Artificial Intelligence modules to provide computerized feedback to trainees. The work reported here consisted of the use of cognitive task analysis methods developed at the University of Idaho to perform knowledge acquisition for a proof of concept training module targeted toward the defensive counter air mission. The specific subtask analyzed was "the use of fire control radar for search and sort" at the beginning of an Air-to-Air intercept performed by F-15 and F-16 pilots. The cognitive task methodology was conceptual graph analysis, a method that uses conceptual graphs to structure interviews and observational data gathering. The analysis consisted of three steps: (1) Development of conceptual graphs from existing documentation; (2) Expansion of the graphs through interviews structured with question probes; and (3) Expansion and completion of the graphs through performance observation and inductive analysis. After the conceptual graph analysis was completed, additional decision heuristics were used to identify the type of expert system architecture(s) most suitable for the task. These architectures include a rule-based system with explanation capability, classifiers with some type of explanation capability, and case-based reasoning with analytical ability