Objective: To review the literature on studies comparing all aspects of problem-based learning with the conventional mode of teaching.
Data sources: Medline literature search (1980 through 1999) and the references cited in retrieved articles.
Data selection: Studies and meta-analyses that compared the newer problem-based learning curriculum and the conventional lecture-based mode of teaching undergraduate medical students. Areas of comparison included the academic process; programme evaluation; academic achievement; graduates' performance, specialty choices, and practice characteristics; and the attitude of students and teachers towards the programmes.
Data extraction: Data were extracted independently by multiple authors.
Data synthesis: Students of the problem-based learning curriculum found learning to be "more stimulating and more humane" and "engaging, difficult, and useful", whereas students of the conventional curriculum found learning to be "non-relevant, passive, and boring". Students who used the problem-based learning method showed better interpersonal skills and psychosocial knowledge, as well as a better attitude towards patients. Students using the conventional model, however, performed better in basic science examinations. Teachers tended to enjoy teaching the newer curriculum. Although the two curricula encourage different ways of learning, there is no convincing evidence of improved learning using the problem-based learning curriculum.
Conclusion: A combination of both the conventional and newer curricula may provide the most effective training for undergraduate medical students.