1. The statistical nature of transmitter release was studied in unparalysed cut rat diaphragm using a voltage clamp technique at room temperature (23 degrees C). 2. While the binomial distribution described observed amplitude histograms well, the Poisson distribution was clearly inadequate. Values of m ranged from 40 to 45, while values for n varied from 45 to 53 and p was between 0.82 and 0.90. 3. Estimation of the probability of release from the transient decay of e.p.c.s in short tetanic trains (11 pulses, 150 Hz) gave values of p from 0.031 to 0.054, which are more than one order of magnitude lower than statistical estimates. 4. As a result of the short tetanic stimulation (5 pulses, 20, 50 and 100 Hz) there is an initial transient facilitation which afterwards becomes masked by depression. Statistical analysis suggests that the changes in the average numbers of quanta released (m) could be attributed to the change in the immediately available store (n). 5. During long tetanic stimulation (4000 pulses, 10-100 Hz) statistical analysis suggests that the decrease in the average number of quanta released (m) could be attributed almost entirely to the decrease in the immediately available store (n). The probability of release (p) decreased only slightly. The extent of the post-tetanic potentiation indicates that it cannot be explained on the grounds of increased probability of release (p) only. There should be an increase as well in the immediately available store (n). 6. It is suggested that while depression is most likely caused by the depletion of the immediately available store due to insufficient replenishment, the facilitation is probably caused by the increase in the capacity of the immediately available store to contain transmitter.