BACKGROUND: Traditional continuing medical education (CME) has not been successful in improving physicians'practice. This project evaluated the use of e-mail to deliver evidence-based moderated case discussions to family physicians. METHODS: In a randomized controlled trial, 58 southwestern Ontario physicians were recruited and randomly assigned to receive two evidence-based cases (type 2 diabetes, prevention) or were put on a waiting list to receive the same. On-line discussions took place about each case. Data were collected using two knowledge questionnaires, charts audits, and standardized patient visits for each of the two cases. RESULTS: The two groups were similar except for rural/urban and solo versus group practice. The latter was related to outcomes, and analyses were controlled for this variable. The intervention group showed statistically significant improvements compared to the control group for knowledge and chart-audit scores for one of the two cases. CONCLUSIONS: Using a randomized control design, this e-mail CME method demonstrated mixed effectiveness.