Portfolio learning has not previously been reported for clinical undergraduate teaching. This open randomized study aimed to assess the effect of portfolio learning in the teaching of oncology to medical students. The project aimed to provide the student with a holistic understanding of the impact of the disease and its treatment on the patient and family, and the natural history of malignant disease, through long-term personal experience of a cancer patient. All undergraduate medical students entering Clinical Studies in October 1992 at the University of Wales College of Medicine were randomized to a study or control group. Both groups continued with the standard curriculum. Each study-group student followed a patient with cancer for 9 months, supported by bimonthly small-group tutorials. Tutors were either general practitioners or hospital consultants, not necessarily oncologists; each was supplied with a tutor's resource pack of key oncology review papers. Students recorded triggers to learning and key items in a personal learning portfolio. Students' performances in clinical examinations and the contents of their portfolio were assessed. Final assessment was by hidden questions in the objective structured clinical examination (OSCE) in the final degree examination, when students in the study group showed higher marks in factual knowledge of oncology, particularly amongst the weaker students (P = 0.01). Those submitting portfolios for formative assessment had higher overall marks than those in the study group who did not (P = 0.04), representing the more motivated students. The whole study group showed a beneficial trend in their knowledge of oncology.