We developed a set of informant-based items describing performance of activities of daily living (ADL) by patients with Alzheimer's disease (AD) to identify which ADL are useful for assessment of patients in clinical trials. Evaluation of ADL is an important outcome measure in AD clinical trials. For clinical trial measurement, ADL should have broad applicability, good test-retest reliability, scaling to cover a range of performance, and sensitivity to detect change in disease progression. A total of 45 ADL items developed from literature review and clinical experience were administered to informants of 242 AD patients and 64 elderly controls as part of the multicenter Alzheimer's Disease Cooperative Study Instrument protocol. Half of the subjects were re-evaluated at 1 and 2 months and all at 6 and 12 months. Controls performed virtually all ADL items optimally at baseline and at 12 months. Among subjects with AD, 27 of the 45 ADL were widely applicable, i.e., performed at baseline or premorbidly by > 90% of subjects; showed good test-retest reliability between baseline and 1 and 2 months; correlated with MMSE scores of AD patients cross-sectionally; and showed a decline in performance from baseline to 12 months in at least 20% of AD patients. ADL could be identified that capture change in functional ability in patients across the entire range of the MMSE. The remaining 18 ADL included several that may be useful for trials that target specific populations, e.g., women with AD. Because change on specific items depends on baseline MMSE, ADL evaluation should include items relevant to the severity of dementia of patients enrolled in a clinical trial.