At some synaptic connections in the central nervous system, amplitude distributions of evoked synaptic currents exhibit surprisingly sharp and regularly spaced peaks. At these connections, detailed analysis of the peaks has led to the proposal that the 'quantal' synaptic current displays very little variability, not only at a release site, but also between release sites. In this study the latter hypothesis has been tested using simulations of evoked transmission. In contrast with previous conclusions, these simulations demonstrate that the experimental observation of regularly spaced peaks in amplitude distributions of synaptic currents is compatible with large underlying differences in the synaptic current amplitudes between release sites. The simulations also reveal that quantal analysis based entirely on the observation and analysis of regularly spaced peaks in evoked synaptic current amplitude distributions, cannot be used with confidence to estimate presynaptic release probabilities, 'quantal' current amplitudes at each release site, or the total number of available release sites. This problem may be a confounding factor in determining whether pre- or postsynaptic changes underlie alterations in synaptic efficacy, such as occurs during long term potentiation.