A 2-year study at the University of Sherbrooke (Quebec) investigated the changes in six medical students' clinical reasoning processes as they participated in a problem-based learning (PBL) curriculum. In each year, students performed a think-aloud protocol with two medical case problems to solve, one in cardiology and one in urology. In the second year of the study, the students were given somewhat more difficult cases to address. Student responses were compared to those of two practicing physicians, neither of them teachers at the institution. Only responses to the urology cases are analyzed here. Results showed that, as anticipated, hypotheses were generated early for both problems, illustrating the students' capacity to transfer the hypothetico-deductive model of reasoning learned in the pre-clinical PBL curriculum. Results also showed that, among the hypotheses generated early, the principal hypothesis was present, similar to the performance of the physician experts, particularly in the less advanced cases. It is concluded that the PBL approach helps students balance the importance they give to case information. The pattern of time taken to consider each case segment was very similar to that of the experts. The two urology cases and data summaries are appended. (Contains 22 references.) (MSE)