Analysis of data from five expert cardiologists indicates substantial agreement on the first step to be taken in response to an emergency case of computer-simulated congenital heart disease. There was little or no agreement among these same subjects on the specific steps undertaken following their initial response. The problem was solved in 11, 10, 8, and 6 steps by four of the cardiologists. The problem-solving sequence employed by one expert resulted in the patient’s death. It is concluded that while the behavior of experts in the hospital and clinic is the primary means of evaluating successful student performance, computer simulations of patient cases offer the opportunity to use expert data in the calibration of student error.