Does volunteer participation reduce depressive symptoms among older people? Does depression influence whether older people will volunteer? Might nonrandom attrition in a longitudinal study bias the relationship between volunteering and depression? This research addresses these questions with three-wave data from the Americans' Changing Lives study. Multigroup structural equation models for complete and incomplete data are used to estimate the joint causal relationship between volunteer work and depression. The results reveal a beneficial effect of formal volunteering on depression, but not for informal helping. Depression was shown to be associated with a subsequent increase in formal volunteering, suggesting voluntarism as a means of compensation. Functional health problems, not depression, emerged as the important barrier to volunteering. In addition, a sample selection effect--depressed persons and nonvolunteers were less likely to complete the panel study--was detected and accounted for in the analysis. The results are discussed in light of the legitimacy of formal social integration. (Contains 3 tables, 1 figure, and 11 notes.)