Background. The criterion validity of the Center for Epidemiological Studies Depression scale (CES-D) was assessed in a group of elderly Dutch community-residents who were self-referred to a prevention program for depression. Methods. Paper-and-pencil administration of the CES-D to 318 elders (55-85 years). Criterion validity was evaluated with the Mini International Neuropsychiatric Interview (MINI), a clinical diagnostic interview based on DSM-IV. Sensitivity and specificity for various cut-off scores of CES-D were compared with the DSM-IV major depressive disorder (MDD) and with clinically relevant depression (CRD), a composite diagnosis of MDD, subthreshold depression or dysthymia. Furthermore the characteristics of true versus false positives were analyzed. Results. For MDD, the optimal cut-off score was 25, (sensitivity 85%, specificity 64%, and positive predicted value of 63%). For CRD, the optimal cut-off was 22 (sensitivity 84%, specificity 60%, and positive predicted value 77%). True positives, MDD and CRD, reported significantly more anxiety symptomatology and more co-morbid anxiety disorders, false positives reported more previous depressive episodes. Conclusions. The criterion validity of the CES-D for MDD and CRD was satisfactory in this semi-clinical sample of elders. Subjects scoring ≥25 constitute a target group for further diagnostic assessment in order to determine appropriate treatment.