OBJECTIVE To review the impact of the clinical education of internal medicine residents on patients' outcomes. METHODS English-language studies of the relation between internal medicine housestaff training and patients' outcomes were systematically identified by a MEDLINE search and from bibliographies and reference lists of recently published articles. RESULTS We hypothesized that the primary impact of internal medicine residency training on patients' outcomes would be the result of: (1) the inexperience of the residents; (2) the heavy workload these inexperienced residents are expected to manage: or (3) some structural feature of the internal medicine teaching services, such as the discontinuity of patient care inherent in night float systems and the fact that residents rotate to different services each month. We also hypothesized that residents may in may ways provide superior care, and many actually improve certain patient outcomes. Housestaff inexperience, workload, and structural features that promote discontinuity have been shown to affect especially outcomes of resource utilization, length of stay, and patient satisfaction. No study has demonstrated that internal medicine residents contribute to excess patient morbidity or mortality. However, the published studies in this area are for the most part retrospective and were conducted 10 to 15 years ago. The full extent of the untoward (or the beneficial) effects of internal medicine residency training on patients' outcomes is unknown. CONCLUSIONS Multisite, prospective studies would remedy the deficiencies in the published research in this area and would yield the most valid insight into the range and extent of the effects of housestaff training on patients' outcomes. In the absence of such studies and in a rapidly changing managed care environment, academic medical centers and departments of medicine need to be aware of those aspects of the clinical education of residents that are most likely to affect patients' outcomes.