The consistent effect of education on volunteering has been explained in a number of ways. In this study we test the hypothesis that perceived control beliefs are partly responsible. Using two waves of panel data from National Survey of Midlife in the United States we estimated cross‐lagged structural equation models in which education is positioned as the exogenous variable and perceived control and volunteering are allowed to be reciprocally related across the two waves. We find that perceived control predicts volunteering, but there is no reciprocal effect: volunteering has no effect on sense of control. One reason, therefore, that educated people are more likely to volunteer is that they have stronger control beliefs. The findings enrich the theory of volunteering by introducing the idea of agency, showing one way in which resources influence the decision to volunteer.