Clinical performance of residents should be assessed as reliably and validly as possible. This study investigated the reliability and validity of an objective structured clinical examination (OSCE) for assessing clinical performance of internal medicine residents. Residents were required to take a 17-patient OSCE in their first and second year. Reliability of the OSCE was 0.40. Validity studies indicated second-year students were significantly better than third-year students for five of six OSCE skill scores; first-year students were significantly better for three scores. Resident's scores for diagnosis, plan, and total significantly increased on their second OSCE. Generally faculty overall ratings of residents' clinical performance did not correlate with OSCE scores. American Board of Internal Medicine certifying examination scores were consistently positively correlated only with diagnosis. This 17-case OSCE is a feasible method for obtaining moderately reliable, valid data not available from other sources about the clinical performance of residents. More cases should be added to increase its reliability.