Stochastic population models are widely used to assess extinction risk under various management scenarios, but due to the lack of independent data, such models are tested only rarely. Here we evaluate the predictive accuracy of a stochastic, spatially explicit, individual-based model of the population dynamics of the Red-cockaded Woodpecker by comparing simulated data with independent empirical data sets from two populations. We examined primary model predictions such as Population size and number of territories, and secondary predictions such as population structure, dispersal success, natal dispersal distances, and age distributions. The model predicted most evaluated parameters with high accuracy, but model performance could be enhanced by including pioneering behavior and by improving estimates of mate and female dispersal behavior. We judge our model to provide reliable predictions when applied to real populations, with a few specific exceptions.