OBJECTIVES To examine a) whether the Geriatric Depression Scale (GDS) can predict clinician-rated suicide ideation and depression, using the 15-, 5-, and 4-item versions, b) whether an additional suicide-ideation item would improve the performance, and c) whether the results vary by age groups. METHODS First-time psychiatric outpatients responded to the GDS. They were subsequently assessed by psychiatrists blind to the GDS, who also indicated whether suicide ideation was present. The performance of the GDS scales was evaluated using receiver operating characteristic curves. Analyses were conducted separately for young-old (aged 60-74 years) and old-old (aged 75 years or older) adults. RESULTS Areas under the curves showed that the different GDS versions were comparable in detecting depression and suicide ideation. For identifying depression, thresholds of 7, 2, and 2 for the 15-, 5-, and 4-item versions were optimal, respectively. In terms of detecting suicide ideation, all measures performed better in old-old than in young-old adults. A single, self-report suicide-ideation item performed better than all multiitem GDS measures. CONCLUSIONS Both the 4- and the 5-item versions are excellent alternatives to the 15-item version, and all are reasonable tools for detecting the presence of suicide ideation also. However, to improve the effectiveness of screening, brief measures of suicide risk should also be included. Even a 1-item measure of suicide ideation can improve clinical decisions tremendously.