The effect of socioeconomic status (SES) on mortality was examined in the community-dwelling elderly. Data were obtained from four population-based studies that enrolled elderly residents of four US communities (East Boston, Massachusetts; New Haven, Connecticut; east-central Iowa; and the Piedmont region of North Carolina) and followed them for 9 years, starting in 1982 or 1986. Higher SES, whether measured by education, by household income, or by occupational prestige, was generally associated with lower mortality. However, the pattern of findings varied by gender and by community. For men, all three SES indicators were associated with mortality in the majority of cohorts. For women, this was true only for income. SES-mortality associations were attenuated but not eliminated after adjustment for behavior and health status. SES-mortality associations were stronger in New Haven and North Carolina than in East Boston and Iowa. The latter communities are more homogeneous with respect to ethnicity, urbanization, and occupational history than the former. Future research should investigate the relative validity of traditional SES measures for men and women and develop more balanced assessment methods. These findings also suggest that it is important to consider not only individual characteristics but also community attributes that mediate or modify the pathways through which socioeconomic conditions may influence health.