Youth who participate in service activities differ from those who do not on a number of key demographic characteristics like socio-economic status and other indicators of risk; and most studies demonstrating positive outcomes among service participants employ small non-representative samples. Thus, there is little evidence as to whether the outcomes associated with service participation are similar among students with varying levels of risk. The National Household Education Survey of 1999, a large nationally representative cross-sectional data set that focused on community service, was analyzed to investigate associations between the risk status of 4,306 adolescent students (50.2% female; 63.3% European American, M age = 15.9), their participation in community service, and their academic adjustment, behavioral problems, and civic knowledge. Because adolescents who participate in service differ from those who do not with respect to demographic characteristics, propensity score analyses were used to correct for potential selection bias in the examination of these relationships. Analyses tested competing theoretical models of service--protective versus compensatory--among students at varying levels of risk, and suggested that service acts as a compensatory factor with respect to academic, behavioral, and civic outcomes. Propensity score analyses revealed patterns suggesting that, in some cases, students with certain demographic profiles that are themselves related to the likelihood of service participation may benefit from service participation more than others. Findings are discussed in terms of their significance for adolescent development, for planning service programs, and for educational policy.