The standard view of the synaptic function in excitatory synapses has been deeply questioned by recent experimental data on hippocampal glutamate synapses both for possible receptor nonsaturation and for larger and non-Gaussian peak amplitude fluctuations. Our previous investigations of the mechanisms involved in the variability of the response of hippocampal glutamatergic synapses, carried out by computer simulation of simple Brownian models of glutamate diffusion, furnished initial evidence about their presynaptic character. A new, refined model, reported here, assumes a collision volume for the glutamate molecule and a more realistic description of receptors and their binding dynamics. Based on this model, conditions for AMPA and NMDA receptor saturation have been investigated and new miniature (or quantal) EPSC parameters have been computed. The results corroborate the hypothesis that the lack of AMPA and NMDA receptor saturation and the EPSC stochastic variability are attributable to the small volume of glutamatergic synaptic vesicles and hence to the small number of glutamate molecules diffusing in the cleft after a vesicle release. The investigations better characterize some not well-known elements of the synaptic structure, such as the fusion pore, and provide useful information on AMPA receptor dynamics. Indeed, a nice fit between computed EPSCs and some miniature EPSCs in recent experimental literature allowed for the computation of new transition time values among the different AMPA receptor states through a trial-and-error optimization procedure. Moreover, the model has been used to evaluate two hypotheses on the genesis of the long-term potentiation phenomenon.