Air Force investigators conducted a three year program to develop a capability for automated quantification and assessment of in-flight pilot performance. Such a capability enhances pilot training by making ratings more objective, valid, reliable and sensitive, and by freeing instructors from rating responsibilities, allowing them to concentrate upon instruction and safety. Researchers implemented instrumentation for recording T-37 flight data and developed hardware and software for measuring pilot performance using recorded flight data. Flight data were then collected and measurement techniques tested, the results of which indicated that automated measurement was feasible. The program also showed that additional data must be acquired for each student, that it is difficult to prepare in-flight data, and that it is necessary to blend automated proficiency ratings with subjective evaluations of students by instructors in order to achieve an optimum evaluation. (PB)