The classical coefficient of variation method for "quantal" analysis of synaptic responses allows unambiguous identification of pre- and postsynaptic loci underlying synaptic plasticity only when extensive simplifying restrictions are made. They include invariance of quantal parameters and the assumption that a single afferent produces the evoked potentials or currents. More general theoretical formulations and simulations demonstrate that the standard criteria do not always provide useful guidelines because when the other sources of physiological variance are included, putative pre- and postsynaptic domains may overlap. For example, data typically interpreted as indicating modifications at both sites can be due to a mechanism localized to only one of the two, if parameter variances are taken into consideration in the case of a single input cell, or if there are multiple inputs and the stimulus does not activate all of them reliably. With this perspective, other physiologically realistic hypotheses relevant to the expression of synaptic plasticity, such as that during long-term potentiation, can be envisioned.