A change from traditional to problem-based learning (PBL) methods in a psychiatry attachment was evaluated by comparing the learning styles, attitudes to psychiatry and examination performance of 2 cohorts of students. It was hypothesised that the PBL curriculum would result in increased deep learning, decreased surface learning, more favourable attitudes to psychiatry and improved examination performance. It was predicted that students' examination success would be related to the use of deep and strategic learning and favourable attitudes. Consecutive cohorts of Year 2 clinical students taught using a traditional psychiatry curriculum (n = 188) and a PBL curriculum (n = 191) were compared. Students completed the Study Process Questionnaire to assess their learning styles and the Attitudes to Psychiatry Scale at the beginning and end of the attachment. Students completed 2 end-of-attachment examinations, a multiple-choice paper and a viva. The PBL curriculum resulted in significantly better examination performance than did the traditional teaching curriculum, both for multiple-choice questions and the viva. No differences in learning styles or attitudes to psychiatry were found between the curricula. Students were significantly more successful in the examinations if they had received the PBL curriculum, were female, and used strategic learning. Examination performance indicated that the PBL curriculum was more successful than the previous course, but that this improvement was not due to students using more effective learning styles or having more favourable attitudes towards psychiatry. It is possible that students learned more effectively during the teaching sessions in the PBL curriculum, but did not change their preferred learning styles.