Does religious involvement influence changes in physical health? We perform a longitudinal analysis of the effect of religious participation on functioning over a 12-year follow-up period, in a large, prospective, representative sample of elderly persons from New Haven, Connecticut, a religiously diverse community. To examine the possibility that disability or changes in disability may be affecting religious involvement, we perform a second longitudinal analysis of changes in religious practices. Finally, we ask whether psychosocial correlates explain the effect of religious involvement on disability. Findings are (a) that attendance at services is a strong predictor of better functioning, even when intermediate changes in functioning are included, (b) that health practices, social ties, and indicators of well-being reduce, but do not eliminate these effects, and (c) that disability has minimal effects on subsequent attendance. The findings illustrate the short- and long-term importance of religious participation to the health and well-being of elderly people, and suggest a particular significance for religious participation in the lives of disabled elders.