The current model does not account adequately for the characteristics of miniature endplate currents (MEPCs). We do not understand their relatively slow rise, the shape of their rise, their variable and sometimes prolonged decay, and the correlation between amplitude and decay time. If we assume that ACh is released from the vesicle through a pore and that the vesicle enlarges as it takes on additional transmitter, the predictions are more like MEPCs. However, previous measurements showed that after quantal size was increased the vesicles in the terminal were not enlarged. This need not be a problem, because some of the ACh is added to vesicles positioned at the active zones, a process known as second-stage loading. By using the false transmitter precursor monoethylcholine we provide additional evidence for second-stage loading. The distribution of quantal sizes at the junction usually does not follow a normal probability distribution; it is skewed to the right. The skew can be accounted for by a model incorporating second-stage loading in which the vesicles are released randomly, without regard to their ACh content. If the vesicles increase in size when they contain more transmitter, only vesicles at the active zone need swell.