The authors compared the continuity of care that family physicians, general internists, and medical subspecialists provided to their adult primary care patients. The 40 study physicians came from large, private multispecialty practices in the San Francisco Bay Area. The three physician types did not differ significantly in the degree of continuity provided, measured by the proportion of total visits to a patient's primary provider (usual provider continuity (UPC) score). Each type provided approximately 80% of its primary care patients' visits. In contrast, the continuity scores of individual physicians ranged widely, from 57% to 98%. Proxy measures of case mix and physician expertise were found to be associated with differing UPC scores. A more detailed exploration of the subspecialists revealed that the lowest levels of continuity were afforded patients with high utilization rates who did not carry a diagnosis in their primary physician's area of subspecialty expertise. The "generalist versus subspecialist" debate assumes that a physician's training background is a major determinant of the quality of primary care delivered. This was not true in this study for the provision of one aspect of quality, a high level of continuity. If factors other than specialty or subspecialty designation are generally found to be the important determinants of continuity, isolated changes in the proportion of physicians receiving generalist versus subspecialty training may have relatively little impact on the level of continuity afforded adult medical patients.