The aims of this study were to determine whether detection of major depression in primary care was associated with improved outcome, and to compare the 4.5 month outcomes of detected and undetected depressed primary care patients and depressed psychiatric patients. Primary care patients with major depression were recruited from the practices of 50 family physicians in Southeastern Michigan using a two-stage selection procedure employing the Center for Epidemiologic Studies-Depression Scale (CES-D) and the Structured Clinical Interview for DSM-III-R (SCID); clinician detection of depression was ascertained by response to a direct query on a rating form. Depressed patients seeking treatment in an outpatient psychiatric setting also received the CES-D and the SCID. Data on patient demographics and clinical characteristics were obtained for both primary care and psychiatric patients. Initial and 4.5 month scores on the Hamilton Depression Rating Scale (HAM-D) were obtained for 34 undetected and 25 detected depressed primary care and 55 depressed psychiatric patients. Improvement in depression over time was assessed by the change in HAM-D scores over the 4.5 months. The three groups did not differ in initial severity. Both psychiatric and undetected primary care patients showed significant improvement at 4.5 months, whereas detected primary care patients did not improve. At 4.5 months there were no differences in mean HAM-D scores between undetected, depressed primary care patients and depressed psychiatric outpatients. This result did not change after controlling for age and severity of depression at initial presentation, nor did it change after exclusion of cases of mild depression to control for a possible "floor effect." However, differences among groups in the stage of depressive episodes may have affected this comparison. These findings suggest that an exclusive focus on increasing detection of depression in primary care patients is unlikely to improve outcomes, and that undetected depression among primary care patients does not necessarily represent poor quality of care. Although depressed psychiatric patients in this study had better outcomes than detected depressed primary care patients, the presence of unmeasured differences among groups in the stage of the depressive episode makes it impossible to determine whether treatment of depression by psychiatrists is superior to that provided by primary care physicians. These findings should stimulate efforts to examine a more comprehensive model for detection and treatment of depression in primary care.