After more than 30 years of experimenting with problem-based learning (PBL) in medical education, the question is still being asked whether PBL is better than the traditional approach in producing competent physicians. Several reviews were published in1993, and then in 2000, some of which challenged the effectiveness and underlying cognitive processing associated with PBL. Others had emphasized the high faculty cost of using PBL and cautioned those who are thinking of using PBL, pointing out the apparent lack of visible difference in the types of graduates from the two streams of teaching and learning. In 1990, a group of medical educations from North America had produced a list of areas where they hypothesized that there would differences between the graduates from a PBL curriculum as compared with those from a conventional curriculum. In this review, I will use the areas as projected by them as the basis, by incorporating results from some of the recent studies not discussed by the reviews from 1993 and 2000. I conclude that differences were indeed found among the graduates from the PBL versus conventional curriculum, and that PBL is still the best method we know to date to train competent physicians.