This research tests the reliability and construct validity of social engagement, a new quality of life measure embedded in the federally mandated Resident Assessment Instrument (RAI) for the nursing home population. The sample consisted of 1,848 residents from 268 homes in 10 states with data collected by trained research nurses. Three resident groups were formed based on residents' cognitive and ADL functioning. Social engagement was significantly related to average time spent in activities across all three groups. We hypothesized a four-factor model with social engagement distinct from mood problems, conflicted relationships, and behavior problems. LISREL confirmatory factor analysis found the data to be consistent with this hypothesis (Fit Index > .98). Intercorrelations between factors showed that for high-functioning residents, engagement was negatively related to conflict but was positively related to conflict among the most impaired. The validity of the social engagement measure and its stability across types of residents suggest its potential utility as a marker of nursing home quality.