To develop, validate, and cross-validate a patient-completed screen for multiple mental disorders in primary care. Comparison of a patient self-report screen with an independent diagnostic assessment by mental health professionals using the Structured Clinical Interview for DSM-III-R diagnoses as criterion standard. Three Rhode Island family practices and a South Carolina family medicine residency. In the initial validation study, 937 patients in Rhode Island were screened; 388 were interviewed. In the cross-validation study, 775 patients were screened in Rhode Island and South Carolina, and 257 were interviewed. SCREEN ITEMS: Sixty-two questions pertaining to nine mental disorders and suicidal ideation. A 16-item screen remained after analysis of item and scale performance. Sensitivity, specificity, and positive predictive value, respectively, were calculated for the following scales: alcohol abuse or dependence (62%, 98%, and 54%), generalized anxiety disorder (90%, 54%, and 5%), major depression (90%, 77%, and 40%), obsessive-compulsive disorder (65%, 73%, and 5%), panic disorder (78%, 80%, and 21%), and suicidal ideation (43%, 91%, and 51%). Replication in a new sample showed attenuated but acceptable operating characteristics for cross-validation. The Symptom-Driven Diagnostic System for Primary Care screen assesses multiple mental disorders that are common to primary care. It serves as a sensitive, valid, and patient-friendly first step in a new approach to recognizing and managing mental disorders in primary care. Finally, it aids the primary care clinician in selecting an appropriate diagnostic interview module for the disease for which the patient screened positive.