Psychometric evaluation of Riemenschneider's (2000) revised Geriatric Depression Scale (RGDS) was undertaken. It was predicted that the RGDS would demonstrate greater sensitivity than the original Geriatric Depression Scale (GDS) in identifying older adults with mild depression. A sample of 72 community-dwelling older adults was invited to complete a structured clinical interview of DSM-IV Major Depressive Episode symptoms (The Mini-International Neuropsychiatric Interview) and fill out a 32-item collapsed self-report inventory that contained RGDS and GDS items. Based on performance on the structured interview, participants were placed in two groups: mild depression (n = 20) and not depressed (n = 52). Results provide adequate evidence of the psychometric properties of the RGDS and GDS. Reliability analyses, utilizing Cronbach alpha coefficients, indicate both measures possess acceptable internal consistency. Receiver operating characteristic (ROC) analysis was performed to assess the discriminate ability of the RGDS and GDS to correctly identify individuals with (test sensitivity) or without (test specificity) mild depression. Area under the curve (AUC) values found both measures demonstrated modest ability to discriminate between groups. As predicted, using suggested cutoff scores, the RGDS demonstrated greater sensitivity than the GDS in correctly identifying individuals with mild depression. However, analysis of positive predictive values (PPV) and negative predictive value (NPV), using exploratory base rates of mild depression, suggest that the GDS is much more precise in interpreting a positive test finding while only somewhat less precise in interpreting a negative test finding. Therefore, the GDS may remain the more clinically useful measure.