BACKGROUND: Older people with depression make greater use of healthcare services, but the detection of the disorder is poor. The National Service Framework for Older People recommended screening for depression in acute healthcare settings to improve health outcomes of older people. Previous studies, mainly outside the UK, report widely differing rates for depression that do not usefully inform UK practice. Thus the aim of this study is to estimate, in a large representative sample of older medical inpatients in a UK hospital setting, the prevalence of depressive symptoms and ICD-10 depressive disorder and to examine the sensitivity and specificity of the 15-item Geriatric Depression Scale (GDS-15) as a screening instrument. METHODS: A two-phase prevalence study of depressive disorder was carried out in acute wards of a district general hospital. Six hundred and eighteen (61%) of 1,009 eligible older medical inpatients were screened using the GDS-15. A stratified sample (n = 223) was further assessed using the Geriatric Mental State, from which ICD-10 diagnoses were determined. RESULTS: The weighted prevalence estimate of ICD-10 depressive disorder was 17.7% (95%CI: 12.9-22.5). Forty-four percent of participants scored above the normally recommended cut-point of >or=5 on the GDS-15. However, on the basis of ROC, the optimal cut-point of the GDS-15 for screening for depressive disorder in this hospitalised population is two points higher at >or=7 (sensitivity 0.74, specificity 0.81). CONCLUSIONS: This study confirms that depression is common amongst older UK medical inpatients with 1 in 6 suffering from clinical depression. The cut-point for GDS-15 for this population is >or=7.