Major depressive disorder is the most common diagnosis encountered in family practice, yet family physicians are relatively unlikely to make the diagnosis. This study compared physician ratings of depression with scores from the Center for Epidemiological Studies-Depression (CES-D) questionnaire and with telephone interview diagnoses of depression using the 3rd revised edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-III-R) criteria for major depressive disorder in a population of 266 patients in community-based family practices. Additional assessments were made of health status, stress, social support, prescribed psychotropic medication, and counseling. The prevalence of positive questionnaire scores in this population was 22.6 percent, and the prevalence of major depressive disorder (based on telephone interview) was 8 percent. Physician ratings of depression were relatively inaccurate when compared with either CES-D scores or telephone interview diagnoses. Optimum specificity (80 percent) and sensitivity (50 percent) with telephone interview diagnoses were achieved when physicians rated the patient as having any depression versus having no depression. Physician ratings of depression were correlated with their assessment of patient stress, social support, and physical health but not with more objective measures of these variables. When compared with telephone interview diagnosis, the sensitivity and specificity of the CES-D scores were relatively poor, suggesting that the CES-D is not useful as a screening tool for unselected populations. Finally, we found that family physicians base their assessments of depression more on distress than on depressive symptoms. Certain physician myths, barriers, and biases may exist that preclude the effective diagnosis of depression.