Since clinical training often fails to equip medical students with essential history-taking skills, more effective teaching methods required to be developed. As previous work had suggested that training which included televised demonstrations of history-taking and practice with simulated patients might be superior to traditional methods, it was decided to evaluate this experimentally. Thirty students beginning the Oxford Clinical Course were randomly selected and allocated to a traditional or one of two experimental courses. Both experimental courses used television and simulation but differed in the way the history-taking demonstrations were presented. After completing these courses, each student was asked to take a history from a simulated patient. Rating of the videotapes of these interviews showed that the two experimental groups obtained much more information and used many more of the required skills than students assigned to the traditional course. The experimental students were also rated more favourably by the simulators and recorded more data in their case histories. It is concluded that these short courses were practical, very effective, and could be augmented by a self-teaching programme in history-taking skills.