Although the use of self-reported ADL (activity of daily living) scales has a long history, the Katz-based assumptions of unidimensionality and hierarchy are increasingly found lacking, and ADLs alone are found to underestimate dysfunction and disability. Data from nearly 8900 elderly respondents in the community sample of the 1991 Canadian Study of Health and Aging were used to examine the measurement properties of a modified version of the Older Americans Research Survey (OARS) ADL and IADL items combined. A multidimensional factor structure was revealed, with three levels of functional ability possessing internal consistency. We conclude that assumptions regarding ADL/IADL unidimensionality and hierarchy are not always valid, and that ADL and IADL items should be considered in combination to capture a greater range of functional disability prevalence. We also suggest that expectations of precise measurement of functional dependence by (I)ADL scales should perhaps be relaxed to the goal of simply differentiating broad levels of self-reported functioning (such as basic, intermediate, and complex), within which some tasks are roughly equivalent. Because these scales are widely used as screening tools and in shaping policy, we suggest that employing a more empirically grounded measurement standard has the potential to reduce bias due to item complexity and task specificity, facilitate standardization, and more reliably predict outcomes.