Use of the Helping Relationship Inventory (HRI) to assess the impact of an introductory medical interviewing course on beginning medical students' preference for interview response modes was studied at the University of Minnesota. The HRI, a brief paper-and-pencil instrument, was administered as a pre-test and post-test for five consecutive years (1975-79). Attention was directed to the impact of the medical interviewing course on students' preferences for the following alternative responses to patients' written statements: "understanding,""probing," interpretive,""supportive," and "evaluative." In 1978 and 1979, the HRI was also administered several momths before the course to assess the persistence of changes. A course description and information on the evaluation instrument and data collection methodology are presented along with pre-test, post-test, and followup test results for each of five consecutive years. Significant increases in students' preferences were found for understanding responses and decreases in preferences for evaluative responses. Changes were in the desired direction, since rapport is generally enhanced by conveying understanding and refraining from premature evaluation. It was also found that tutor experience in teaching the course was not significantly related to variations in outcome results; students with new tutors did not have significantly different response preferences than students of more experienced tutors. The consistent pattern of results for five years supports the construct validity of the HRI. Effects on response preferences of some instructor characteristics are analyzed, and implications for health professional education and research are discussed. (SW)