Older residents (N 5 1972) in California were investigated prospectively for association of volunteering service to others and all-cause mortality. Potential confounding factors were studied: demographics, health status, physical functioning, health habits, social support, religious involvement, and emotional states. Possible interaction effects of volunteering with religious involvement and social support were also explored. Results showed that 31 percent (n 5 630) of respondents volunteered, about half (n5289) for more than one organization. High volunteers ([.greaterequal]2 organizations) had 63 percent lower mortality than nonvolunteers (age and sex-adjusted) with relative hazard (RH) 5 0.37, confidence interval (CI) 5 0.24, 0.58. Multivariate adjustment moderately reduced difference to 44 percent (RH 5 0.56, CI 5 0.35, 0.89), mostly due to physical functioning, health habits, and social support. Unexpectedly, volunteering was slightly more protective for those with high religious involvement and perceived social support. After multivariate adjustment, any level of volunteering reduced mortality by 60 percent among weekly attenders at religious services (RH 5 0.40; CI 5 0.21,0.74). Lower mortality rates for community service volunteers were only partly explained by health habits, physical functioning, religious attendance, and social support.