The sense and access optimization problem in opportunistic spectrum access technology is considered. By introducing the concept of event, the original problem with countably infinite states is converted to an event-based decision problem with finite events. From a sensitivity-based view, the performance differences between two event-based policies are investigated. Based on the event-based performance difference formula, with the appropriate approximation, the sample-path based policy iteration algorithm is developed. An example is provided to illustrate the effectiveness and reasonableness of the proposed algorithm.