A study of third-year medical school discussion groups was undertaken to determine how much the cognitive level of instructors' questions in small group sessions influenced student responses and whether these responses had any measureable relationship to critical thinking skills, or National Board of Medical Examiner (NBME) scores. The research model was based on concepts of process-outcomes designs, and the emphasis was to relate the verbal participation of the individual student to certain outcomes thought to be important to students in clinical training. The study sample consisted of 110 third-year medical students and 22 medical faculty. Students were assessed using the Medical College Admission Test, NBME preclinical performance appraisal, the Watson Glaser Critical Thinking Appraisal Test A, and three scales of the California Personality Inventory. The observational tool for the process measures was a modified version of the Flanders System of Interaction Analysis. Both teacher questions and student discussion incorporated the congitive levels of Bloom's taxonomy. Two major study findings were: (1) the cognitive level of the teacher's questions was significantly correlated with the cognitive level of the individual student's response, particularly at the higher cognitive levels of analysis (synthesis and evaluation); and (2) the student's entry characteristics had a far larger effect on outcome measures than did the process of talk variables in the small discussion groups. (SW)