This paper develops a dynamic model of voluntary affiliation which describes the flow of individuals into and out of organizations over time. The model is a simple stochastic process model, developed from Coleman, which we use as a baseline generator to produce specific predictions for the distribution of affiliation in cross-section. Incorporating well-known empirical generalizations, the model produces some surprising results for the well-known class differences in the rate of affiliation. The model is fitted in cross-section for six countries, and estimates of the dynamic parameters are produced from a panel study in the United States. We discover some systematic negative consequences for low status individuals and organizations, and some surprisingly strong support for the basic imagery of our model. We conclude with an exploration of several consequences, both substantive and methodological.