Background: Decline in physical functional ability is an intrinsic component of the dementia syndrome. Reductions in muscle mass and strength represent a major factor in the loss of functional ability. Although resistance exercise has been studied as a method for maintaining/recovering function in populations of frail older adults, people with dementia have been systematically excluded because of uncertainty about the reliability of outcome measurements. Objective: The purpose of this study was to determine the test-retest reliability of a battery of established performance-based measures of strength and function among subjects with dementia. Setting: A hospital-affiliated adult day-care facility. Subjects: Twelve of 28 older subjects with dementia of various etiologies were available for two assessments prior to implementation of a resistance-exercise intervention. Methods: Subjects underwent an assessment of lower extremity strength and physical function consisting of two recorded trials of bilateral isometric strength of the knee extensor, hip flexor, and dorsiflexor muscles, as well as hand grip strength; repeated chair stands, evaluation of usual- and maximal safe-gait speed over a 6-m course, and the Timed-Up-and-Go Test. The entire assessment was repeated approximately 1 week later. An average of the trials for each measurement was computed for each of the two assessment periods, and intraclass correlation coefficients (ICCs) for these paired measurements were estimated using STATA. Results: ICCs ranged from .56 for left iliopsoas to .77 for left dorsiflexors among the strength measures whereas measures of function ranged from .80 for number of steps in usual gait to .95 for time of fast gait. Conclusions: Performance-based measures of strength and function can be reliably assessed in older people with dementia, although measures of function appear to be more reliable than measures of strength.