This paper explores the causal effect of volunteer work providing daily assistance to the elderly on elderly mortality. To identify the causal effect, I exploit the earthquake that occurred in midwestern Japan in 1995 as a natural experiment inducing exogenous variation in the level of volunteering. The municipalities hit by the earthquake experienced a sharp increase in the level of volunteering. Based on a comparison of mortality between the municipalities with no or little loss of life due to the earthquake, that experienced the sharp increase in the level of volunteering, and the nearby municipalities that were not hit by the earthquake, I find that volunteering significantly reduced elderly mortality. Close attention is paid to ensure that the results are not driven by the direct effects of the earthquake.