The psychometric properties of the Beck Depression Inventory and its short form were compared for a psychiatric sample of clinically depressed adults. Factor analyses indicated that the short form retained the essential integrity of cognitive components of the standard inventory but further attenuated already weak somatic components. As a result, the wisdom of shortening the scale is questioned and a call is made for improving the existing scale. A preliminary attempt at improvement, involving the modification of items inconsistent with other diagnostic systems, proved reasonably successful. Further suggestions for enhancing the psychometric properties of the scale are discussed.