A Monte Carlo method for modeling the neuromuscular junction is described in which the three-dimensional structure of the synapse can be specified. Complexities can be introduced into the acetylcholine kinetic model used with only a small increase in computing time. The Monte Carlo technique is shown to be superior to differential equation modeling methods (although less accurate) if a three-dimensional representation of synaptic geometry is desired. The conceptual development of the model is presented and the accuracy estimated. The consequences of manipulations such as varying the spacing of secondary synaptic folds or that between the release of multiple quantal packets of acetylcholine, are also presented. Increasing the spacing between folds increases peak current. Decreased spacing of adjacent quantal release sites increases the potentiation of peak current.