Problem-oriented case discussion is a valuable teaching tool in clinical pharmacology. Small groups are ideal for discussion but attempts to extend this to large groups often magnify the problems seen with small group discussion to an unacceptable extent. We describe a format which overcomes many of the problems encountered. The class is divided into groups of six students. Cases for discussion are distributed a week before discussion. Individual groups meet, research, discuss, form a consensus opinion and appoint a chairperson for each topic. The class meets and two groups are randomly chosen to discuss the first topic. The chairperson of the first group briefly presents the group's analysis of the problem. The chairperson of the second group then acts as an ‘interrogator’ questioning statements of fact, application of knowledge or logic, and offers alternative opinions. After discussion between the two groups the rest of the class join the discussion. Two new groups are chosen for the second case. This format facilitates self-directed learning, well-prepared participants and worthwhile discussion.