Several potential advantages for students' learning are claimed for problem-based learning (PBL). Students in PBL curricula may be more highly motivated; they may be better problem solvers and self-directed learners; they may be better able to learn and recall information; and they may be better able to integrate basic science knowledge into the solutions of clinical problems. Although some of these claims find theoretical support from the literature on the psychology of learning, to date there has been no review of the experimental evidence supporting the possible differences in students' learning that can be attributed to PBL. In this review article, the authors examine each claim critically in light of that evidence. They conclude that (1) there is no evidence that PBL curricula result in any improvement in general, content-free problem-solving skills; (2) learning in a PBL format may initially reduce levels of learning but may foster, over periods up to several years, increased retention of knowledge; (3) some preliminary evidence suggests that PBL curricula may enhance both transfer of concepts to new problems and integration of basic science concepts into clinical problems; (4) PBL enhances intrinsic interest in the subject matter; and (5) PBL appears to enhance self-directed learning skills, and this enhancement may be maintained.