BACKGROUND: The Youth Self-Report (YSR) has been used widely as a screening instrument for adolescent psychopathology. The present study aimed at a test of the diagnostic accuracy of the various YSR-scales including a DSM-oriented affective problem scale (YSR AFF) in the prediction of depressive episodes and a comparison with results based on the Center of Epidemiologic Studies-Depression Scale (CES-D). METHODS: A consecutive clinical sample of 140 adolescents diagnosed with major depressive episodes according to ICD-10 criteria was compared to a sample of 140 non-referred controls matched by age and sex from a community survey. All subjects responded both to the YSR and CES-D. Diagnoses were provided by the treating clinicians. Receiver Operating Characteristics (ROC) analyses were performed and cut-off scores were calculated based on quality efficiency statistics. RESULTS: The YSR AFF scale was found to have high diagnostic accuracy and showed quite comparable results to the CES-D scale. None of the other multivariate model showed a better performance in the identification of major depression disorders. Based on quality efficiency indicator analyses, scores between 5 and 9 on the YSR AFF scale and between 12 and 31 on the CES-D scale served best in the prediction of clinical depressive episodes in adolescents. LIMITATIONS: No formal reliability test of the diagnoses was available. CONCLUSION: The DSM-oriented YSR AFF scale shows a high diagnostic accuracy and can be recommended for the clinical assessment of depression in adolescents.