The purpose of this study was to examine the Geriatric Depression Scale (GDS) translated into Turkish for its reliability, discriminant validity, and factor structure in a sample of 276 community-dwelling elderly and 30 patients with major depression. One item (Item 5) was discovered to have conceptual difficulty for Turkish elderly and was transformed to negative form. Item 2 was transformed to positive form to keep the number of positive and negative items equal to that in the original GDS. A reasonable time stability with 1-week interval (r:.74) and a high level of internal consistency (alpha = .91) were observed. Student's t test resulted in a significant discriminant validity for the scale total score. Factor study with principal component analysis and varimax rotation gave rise to a structure with seven factors. Results of the same analysis with two factors were found to be easier to interpret. The first factor was composed of 19 items reflecting "depressive affect and thought content." The other 11 items representing "decrease in motivation and cognitive functions" loaded in the second factor. In conclusion, the Turkish GDS was found to have reasonable time reliability, high internal consistency, and discriminant validity for Turkish elderly. Its two-factor structure can be used as an informative instrument for epidemiological studies, reflecting two main dimensions of depression in the elderly.