BACKGROUND: A 15-item GDS form is now widely used to screen for depressive symptoms in elderly patients, but even the 15-item version seems fairly long for routine use in busy outpatient settings. This study was conducted to determine which of the three short-form GDS tools with four or five questions extracted from the 15-item GDS tool could be a valid alternative to the 15-item GDS. METHODS: This study was performed using a questionnaire survey of 102 elderly patients over sixty-five, who had either visited or had been admitted in either one of the two universities hospitals in Seoul or ChungCheong district from October 1999 to June 2004. Questionnaire contained demographic and statistical characteristics, items related to health behavior, 15-item GDS, Mini-Mental State Examination Korean (MMSE-K) and activities of daily living. Three scales, namely, D'Ath GDS-4, Van Marwijk GDS-4, and Hoyl GDS-5 were compared to the 15-item GDS. Sensitivity, specificity, positive predictive value and the area under the ROC curves were calculated for each short-form GDS scales. RESULTS: The mean 15-item GDS score was 7.4+/-3.3 and 69.6% of the subjects were to have suggested depression. Hoyl GDS-5 instrument had the highest sensitivity of 87.3%, specificity of 80.6%, and positive predictive value of 91.1% among the three short-form alternatives. The 5-item Hoyl GDS version had the highest area under the curve (0.84), compared to 0.76 for the 4-item Van Marwijk and 0.66 for the 4-item D'Ath GDS version. CONCLUSION: The 5-item Hoyl GDS had superior validity over the D'Ath GDS-4 and Van Marwijk GDS-4. The GDS5/15 may be an effective alternative screening tool.