This study looked at several aspects of clinical instruction in small groups such as teacher questioning and student discussion and documented how certain student cognitive outcomes are related to them. Drawing from actual transcripts of over 75 hours in small group sessions, the study examined in detail such things as the logic and level of the questions medical teachers ask, the quality of student responses, the cognitive level of classroom discourse and the effects of instruction on selected outcomes such as the students' ability to think critically and their performance on National Board of Medical Examiners examinations. The instrument used for teaching analysis was a modified version of the Flanders System of Interaction Analysis which incorporated Bloom's Taxonomy of the Cognitive Domain. The cognitive level of the teachers' questions had a significant relationship to the cognitive level of the students' response and discussion. While the students who talked at the highest cognitive level in class were those who scored highest in National Board Examinations (Parts 1 and 2) and the critical thinking tests, the multiple regression analyses showed that classroom activity contributed little to outcomes. The major contributions to outcomes were the aptitude and previous knowledge that the students already possessed.