AVL systems give the dispatcher of emergency units (e.g., police cars, ambulances) the estimated real time locations of units within each service area. AVL cost is projected to range from $500 to over $10,000 per vehicle. Building on a recently developed hypercube queuing model, this paper presents a Markov process model for computing the operating characteristics of the radio-dispatched fleet under a policy whereby the closest available unit is dispatched to each call for service. The model accommodates a realistic description of the service area and general spatial deployment policies for units. A realistic nine-unit police example is provided that indicates the general ways in which AVL dispatching affects system performance. The model thus allows a detailed depiction of a city's geography and travel time characteristics, as well as flexible procedures for deploying patrol units. Using the model, a potential consumer, manufacturer, or evaluator of an AVL system can estimate the AVL effect on response times, interarea dispatch frequencies, workloads, dispatch error probabilities, and other measures of system performance. Tabular data, footnotes, 14 references, and a discussion of computational work are provided. (Author abstract modified)