Depression in oncology patients is under-recognised and associated with poor outcomes. Screening can increase case recognition. The Brief Case-Find for Depression (BCD) is a four-question, interviewer-administered instrument that has been previously validated in a general medical setting. The primary aim of this study was to validate the BCD in a medical oncology/palliative care setting, primarily by comparing its association with physical illness measures and with the Primary Care Evaluation of Mental Disorders (PRIME-MD), the Beck Depression Inventory (BDI) and the Hospital Anxiety and Depression Scale (HADS). Eligible adult oncology patients gave informed consent and completed the above measures and a pain scale. Agreement between the BCD and other instruments was assessed. Construct validity was determined by comparing depressed/nondepressed patients with respect to performance status, symptoms, pain score and functional impairment. A total of 100 patients had a median age of 58 (range 21-90) and ECOG performance status (PS) 2 (0-4). In all, 60% had metastatic disease. The therapeutic goal was curative/adjuvant in 39% and palliative in 61%. Prevalence of depression according to the various measures was: BCD 34%, PRIME-MD 12%, BDI 19% and HADS 14%. In total, 45% of patients responded positively to a single-item screening question. The BCD showed fair agreement with the PRIME-MD (kappa=0.21), moderate agreement with the BDI (kappa=0.43) and fair agreement with the HADS (kappa=0.27). Against the PRIME-MD diagnosis of depression, the BCD had greater sensitivity, but lesser specificity and overall agreement, compared with the BDI and depression scale of the HADS. Patients with probable depression (according to BCD) had inferior PS (P=0.0064), increased pain (P=0.045) and greater impairment of functioning (PRIME-MD: P=0.0003). There was no association with gender, age, disease status or therapeutic goal. Depression is common in oncology patients. The BCD is a quick, easy-to-administer screen for depression, which has reasonable psychometric properties in this population.