A Physical Activity Scale for the Elderly (PASE) was evaluated in a sample of community-dwelling, older adults. Respondents were randomly assigned to complete the PASE by mail or telephone before or after a home visit assessment. Item weights for the PASE were derived by regressing a physical activity principal component score on responses to the PASE. The component score was based on 3-day motion sensor counts, a 3-day physical activity dairy and a global activity self-assessment. Test-retest reliability, assessed over a 3-7 week interval, was 0.75 (95% CI = 0.69-0.80). Reliability for mail administration (r = 0.84) was higher than for telephone administration (r = 0.68). Construct validity was established by correlating PASE scores with health status and physiologic measures. As hypothesized, PASE scores were positively associated with grip strength (r = 0.37), static balance (r = +0.33), leg strength (r = 0.25) and negatively correlated with resting heart rate (r = -0.13), age (r = -0.34) and perceived health status (r = -0.34); and overall Sickness Impact Profile score (r = -0.42). The PASE is a brief, easily scored, reliable and valid instrument for the assessment of physical activity in epidemiologic studies of older people.