OBJECTIVE To evaluate prospectively the reliability and validity of the Geriatric Depression Scale administered by telephone (T-GDS) in patients undergoing outpatient comprehensive geriatric assessment. METHODS A total of 101 geriatric patients were evaluated in a 1-year period at the outpatient Geriatric Assessment Center of the University of Nebraska Medical Center. METHODS The 30-item GDS was completed by all patients on three occasions: by telephone several days before their assessment, face-to-face during their assessment visit, and several days later, again by phone. During their assessment, all patients were evaluated by one of three geriatric psychiatrists who were blind to all GDS results. The test-retest reliability of the T-GDS was measured by comparing the results of the two phone interviews. The construct validity of the T-GDS was estimated by comparing the results of the initial T-GDS to the GDS obtained during the comprehensive assessment. The criterion validity of the T-GDS was estimated by comparing the results of the T-GDS with the clinical diagnosis of depression assigned by the psychiatrists. RESULTS The individual items of the initial T-GDS showed substantial concordance with the second T-GDS (kappa range 0.35-0.75, mean = 0.52), and with the assessment GDS (kappa range 0.29-0.75, mean = 0.52). One item showed evidence of bias when comparing the two T-GDSs, and two items when comparing the initial T-GDS to the GDS done during the assessment. The mean number of symptomatic responses was not significantly different for the T-GDS versus assessment administration but did decline slightly when comparing the two T-GDSs. ROC curve analyses showed good agreement between the clinical diagnosis and the T-GDS. CONCLUSIONS The GDS appears to maintain its reliability and validity when administered via telephone and thus may be useful for a variety of epidemiological and clinical purposes.