The reliability of the Functional Assessment Inventory (FAI) was evaluated using a sample of VA domiciliary and nursing home patients. The interobserver and interrater reliability coefficients of the summary rating scales, based on a single assessment, tended to be higher than their test-retest reliability coefficients, based on two independent assessments separated by a modal four-week interval. Validity coefficients, using the OARS instrument ratings as criteria, also based on two independent assessments several weeks apart, were, on the average, as high as the test-retest reliability coefficients. More specifically, the mental health, physical health, and activities of daily living rating scales, along with the objectively scored Short Portable Mental Status Questionnaire and Short Psychiatric Evaluation Schedule, tended to yield relatively similar scores with repeated measurement, while the social resources and economic resources scales were somewhat less stable, a discrepancy possibly explained by the homogeneous nature of the social and economic status of most of the patients (institutionalized veterans). Thus the reliability and validity of the FAI are satisfactory, but the stability of some of its scales requires further investigation.