PURPOSE: The purpose of this study was to validate an item bank of questions associated with activities of daily living (ADL), using graded item response theory (G-IRT), in a sample of community-dwelling survivors of a stroke. METHODS: The study sample consisted of 364 community-dwelling individuals who reported a recent history of stroke. Sixteen line items from the 1999-2000 National Health and Nutrition Examination Surveys (NHANES) survey were analyzed using factor analysis, internal analyses of consistency, and G-IRT. RESULTS: The 16 line items demonstrated unidimensionality and were internally reliable. Thirteen line items demonstrated good discrimination and suitable thresholds. The majority of items exhibited appropriate sensitivity across the entire spectrum of functional severity indicating that these items closely reflected the relationship of decreased function with increased severity of illness. CONCLUSION: Each of the 13 items is scale independent, is valid for measurement of functional impairment, and may be applicable for use in a scale for assessment of functional change in community-dwelling stroke survivors.