We report findings from an intervention study that investigates the impact of group reminiscence (GR) and individual reminiscence (IR) activities on older adults living in care settings. This research aimed to provide a theory-driven evaluation of reminiscence based on a social identity framework. This framework predicts better health outcomes for group-based interventions as a result of their capacity to create a sense of shared social identification among participants. A total of 73 residents, living in either standard or specialized (i.e., dementia) care units, were randomly assigned to one of three interventions: GR (n = 29), IR (n = 24), and a group control activity (n = 20). The intervention took place over 6 weeks, and cognitive screening and well-being measures were administered both pre- and post-intervention. Results indicated that only the group interventions produced effective outcomes and that these differed as a modality-specific function of condition: Collective recollection of past memories enhanced memory performance, and engaging in a shared social activity enhanced well-being. Theoretically, these findings point to the important role that group membership plays in maintaining and promoting health and well-being.