A radioiodinated alpha-scorpion toxin (toxin II from Androctonus australis Hector) (alpha ScTx) was used as a probe for EM autoradiography to study the distribution of voltage-dependent sodium channels (Na+ channel) on the postsynaptic side of the mouse neuromuscular junction. Silver grain distribution was analyzed by the cross-fire method to assess the relative Na+ channel density in each membrane domain measured by stereology. This analysis showed that the maximum Na+ channel density was located on the edge of the synaptic gutter, where it reached about twice the mean density in the postsynaptic fold membrane. Na+ channel densities have been calculated using ACh receptor (AChR) density in fold crests as reference. Sodium channel density on the edge of the synaptic gutter was estimated at about 5000/microns 2. Sodium channel distribution in the postsynaptic folds was compared to AChR distribution using density distribution analysis (Fertuck and Salpeter, 1976). The results confirmed that, as already observed by immunogold labeling (Flucher and Daniels, 1989), there are no Na+ channels on fold crests. Na+ channels are located in the rest of the fold membrane (bottom) and may be distributed according to two possible models. In the first, density would be uniformly high, although lower than on the gutter edge. In the second, density would decrease from the crest border, where the value was that of the gutter edge, to the fold end, where the value would be 50% lower. Based on the latter model, which was the "best-fit model," we propose that the postsynaptic membrane includes two domains. The first is the fold crest, which contains almost exclusively AChRs. This domain is devoted to reception-transduction of the chemical signal. The second includes both the fold bottom membrane and the perisynaptic membrane. Sodium channel density is highest along the crest border and decreases moving away. Its functions are the integration of postsynaptic potentials and generation-conduction of the muscle action potential.