We investigated the reliability, using a telephone interview procedure, of cognitive, functional, and behavioral scales in an elderly population with normal aging and dementia. Two clinicians performed the assessments: one performed the assessments in a telephone interview format and the other conducted the assessments at the clinic. The telephone interview always preceded the clinic evaluation (2-30 days apart), and both clinicians were blind to any previous evaluations of the patient. The intraclass correlation coefficients between the telephone interview and the ratings obtained by a different clinician on the clinic evaluation varied between 0.92 and 0.98 (P's < or = .001) for comprehensive test scores. These results indicate that a telephone interview format, although not a substitute for a face-to-face diagnostic evaluation, is a reliable procedure for obtaining the assessment modalities studied. These findings are particularly important in aged and dementia research populations where personal contact may not always be feasible.