Abstract An operational two-stage expert system is built with rule induction. The first stage examines the admission decision process for applicants to an MBA program, while the second stage focuses on the prognosis for degree completion for those actually admitted. It is this performance prediction capability that is submitted as the major contribution of the system. While given the opportunity to make use of personal demographic variables that would be suggestive of discriminatory academic policies, the system's pattern recognition algorithm established an optimal rule structure based solely upon academic and professional backgrounds. This induced rule structure was found to be consistent with all cases in the training subset; the rules were then validated on an independent hold out sample.