We analyzed population-based data for respondents aged 70 and older from the Longitudinal Study on Aging. We compared mortality risk among 358 baseline-year working with 4,373 nonworking respondents. Including other demographic and health status predictors in a multiple logistic regression model, employment remained a significant predictor of survival (mortality odds ratio (OR) = 0.4). Diabetes mellitus was the only factor that displayed a mortality risk that was significantly interactive with work (OR = 3.5). These data suggest that a healthy worker effect persists in older age groups, but that within the working stratum, patterns of risk may differ as compared to those among the nonworking.