BACKGROUND Several studies have described etiological and clinical differences between elderly depressed patients with early onset of their illness compared to late onset. While most studies have been carried out in clinical samples it is unclear whether the findings can be generalized to the elderly population as a whole. The aim of this study was to compare early-onset (EOD) and late-onset (LOD) depressive illness in a community-based sample. METHODS Large (n = 3107) representative sample of older persons (55-85 years) in the Netherlands. Two-stage screen procedure to identify elderly with MDD. The Center for Epidemiologic Studies Depression scale (CES-D) was used as a screen and the Diagnostic Interview Schedule (DIS) to diagnose MDD. Data on 90 older persons with early-onset depression and 39 with late-onset depression were available. RESULTS Those with LOD were older, and more often widowed. Family psychiatric history, vascular pathology, and stressful early and late life events did not differ between groups. EOD subjects had more often double depression and more anxiety. CONCLUSIONS In a community-based sample we did not detect clear differences in etiology and phenomenology between EOD and LOD. This discrepancy with reports from clinical samples could be due to selection bias in clinical studies. Consequently, all patients with late-life depression deserve a diagnostic work-up of both psychosocial and somatic risk factors and treatment interventions should be focused accordingly.