Self-report measures of depressive symptoms have been frequently employed in medically ill samples despite the lack of information regarding their psychometric properties in these subjects. We have examined the agreement of the Beck Depression Inventory (BDI) with DSM-III diagnoses of depression in a sample of renal dialysis subjects (n = 99). A threshold of greater than or equal to 15 on the BDI produced optimal sensitivity (.92), negative predictive value (.99), and maximized Youden's index of validity (.72) for the use of the BDI as a screening device for depressive syndromes in dialysis patients. Higher BDI threshold levels decreased the sensitivity of the measure but did not produce a clinically meaningful increase in the positive predictive value.