The Geriatric Depression Scale (GDS) exists in both short and long forms. The original 30-item form of the GDS has been shown to be an effective screening test for depression in a variety of settings. However, its utility in patients with dementia of the Alzheimer type (DAT) is questionable. The short, 15-item version of the GDS was developed primarily for brevity and, in particular, for use in populations such as the medically ill or those with dementia, where the longer form might be burdensome. How well this short form works in these populations, however, is largely undetermined. In this paper, the sensitivity and specificity of the 15- and 30-item GDS are compared in a group of patients who were either cognitively intact or had mild DAT. The findings suggest that the short version of the GDS, like its longer predecessor, is an effective screening tool in the cognitively intact. However, in a population of subjects with mild DAT, it does not appear to retain its validity.