A theory of trading volume is developed based on assumptions that market agents frequently revise their demand prices and randomly encounter potential trading partners. The model describes two distinct ways informational events affect trading volume. One is consistent with conjectures made by empirical researchers that investor disagreement leads to increased trading. But the observation of abnormal trading volume does not necessarily imply disagreement, and volume can increase even if investors interpret the information identically, if they also have had divergent prior expectations. Simulation tests support the model and are used to contrast the random-pairing environment with costless market clearing. Volume is lower in the costly market, and volume increases caused by an informational event persist after the event period. This is consistent with existing empirical evidence and suggests that markets do not immediately clear all orders or that investors have demands to recontract.