1. Following motor nerve stimulation there is a period of greatly enhanced quantal release, called the early release period or ERP (Barrett & Stevens, 1972b). Until now, measurements of the probability of quantal releases at different points in the ERP have come from experiments in which quantal output was greatly reduced, so that the time of release of individual quanta could be detected or so that the latency to the release of the first quantum could be measured. 2. A method has been developed to estimate the timing of quantal release during the ERP that can be used at much higher levels of quantal output. The assumption is made that each quantal release generates an end-plate current (EPC) that rises instantaneously and then decays exponentially. The peak amplitude of the quantal currents and the time constant for their decay are measured from miniature end-plate currents (MEPCs). Then a number of EPCs are averaged, and the times of release of the individual quanta during the ERP estimated by a simple mathematical method for deconvolution derived by Cohen, Van der Kloot & Attwell (1981). 3. The deconvolution method was tested using data from preparations in high-Mg2+ low-Ca2+ solution. One test was to reconstitute the averaged EPCs from the estimated times of quantal release and the quantal currents, by using Fourier convolution. The reconstructions fit well to the originals. 4. Reconstructions were also made from averaged MEPCs which do not rise instantaneously and the estimated times of quantal release.(ABSTRACT TRUNCATED AT 250 WORDS)