The effects of sociodemographic variables on mortality for U.S. adults are estimated, using data from the National Health Interview Survey matched with National Death Index data. Dependent variables of the proportional hazards models include all-cause mortality and mortality due to circulatory disease, lung cancer, breast cancer, other cancer, other medical causes, and nonmedical causes. Results indicate that all of the sociodemographic variables (education, income, employment status, marital status, race) have independent effects on all-cause mortality, with similar-effect sizes for males and females, and weaker effects in the older group (ages 65+) than the younger (ages 25-64). For the separate causes of death, the effects of most sociodemographic variables vary by sex.