Seventeen randomly selected clinical teaching sessions in a medical school clerkship were videotaped during a one-month period, including teaching rounds, working rounds, morning report sessions, lectures, patient management conferences, grand rounds, and journal clubs. Using a verbal behavior classification schedule, the investigators analyzed the videotapes in terms of the proportion of talking done by clerkship instructors, medical students, residents, and others. The nature of the verbal interaction was examined by assessing the proportion of time devoted to giving information versus asking questions. The proportions on these dimensions were further analyzed according to the cognitive level of the verbal interaction. The data suggest that the teaching observed was not optimal for promoting problem-solving ability, since students were placed in a very passive role in which they received a preponderance of low-level, factual information.