PURPOSE: There is little empirical translation of multimodal cognitive activity programs in "real-world" community-based settings. This study sought to demonstrate in a short-term pilot randomized trial that such an activity program improves components of cognition critical to independent function among sedentary older adults at greatest risk. DESIGN AND METHODS: We randomized 149 older adults to Experience Corps (EC) or a wait-list control arm. Participants randomized to EC trained in teams to help elementary school children with reading achievement, library support, and classroom behavior for 15 hr/week during an academic year. We compared baseline and follow-up assessments of memory, executive function (EF), and psychomotor speed at 4 to 8 months by intervention arm, adjusting for exposure duration. We observed a range of EF abilities at baseline and stratified analyses according to the presence of baseline impairment using established norms. RESULTS: Overall, EC participants tended to show improvements in EF and memory relative to matched controls (ps < .10). EC participants with impaired baseline EF showed the greatest improvements, between 44% and 51% in EF and memory at follow-up, compared to declines among impaired-EF controls (ps < .05). IMPLICATIONS: Short-term participation in this community-based program designed to increase cognitive and physical activity in a social, real-world setting may train memory and, particularly, executive functions important to functional independence. This community-based program represents one potentially effective model to bring high doses of sustainable cognitive exercise to the greatest proportion of older adults, particularly those sedentary individuals at elevated risk for health disparities.