The principal intent of this paper is to replicate formally the work of Lemon, Bengtson, and Peterson (1972), the first researchers to carefully articulate a systematic statement of the activity theory of aging, but whose test of the theory resulted in disappointingly ambiguous findings. Probability samples of three distinct types of retirement communities (N = 1209) provided far greater variation in background variables than existed in the original study. Behaviorally based activity scales were drawn from daily activity inventories rather than the single-item ordinal measures of activity items used in the original research. Multiple Classification Analysis was employed to sort out the effects of each activity type from the others and from the effects of age, gender, and health upon life satisfaction. Different from the results of Lemon, Bengtson, and Peterson, our findings, although mixed, lend strong support to the activity theory of aging. Informal activity contributed positively, strongly, and frequently to the life satisfaction of respondents. Solitary activities had no effect on life satisfaction. Formal activity had a negative effect. All activity effects were similar in the three communities. The implications of these findings for activity theory are discussed at length.