This paper describes the extent of release and terminal variability among normal frog sartorius neuromuscular junctions and seeks physiological correlates for these differences. Terminal length varied over approximately a 10-fold range, quantal content and release per unit terminal length ("release efficacy") over much larger ranges. For purposes of comparison of different junctions, release efficacy in a Ringer's containing 0.25 mM Ca2+ was determined in all cases. In a Ringer's containing 0.1 mM Ca2+, tetanic stimulation causes a buildup of evoked release and of miniature endplate potential (mEPP) frequency. The mEPP frequency at the end of the tetanus is proportional to the evoked release level. Following the tetanus, the mEPP frequency declines in a multiexponential fashion, with the 2 longest decay phases, representing augmentation and posttetanic potentiation (PTP), both having time constants that are positively linearly correlated with the synaptic release efficacy. Longer or higher-frequency tetanic stimulation resulted in a longer time course of decay of mEPP frequency. In a Ca2(+)-free/EGTA Ringer's, tetanic stimulation causes no evoked release, but does lead to an increased mEPP frequency, presumably due to a buildup of free Ca2+ displaced from internal stores by the Na+ influx. Following the tetanus, the mEPP frequency declines to resting level with a time constant that is essentially the same for all junctions, regardless of their release efficacy in Ca2(+)-containing Ringer's. These findings indicate that stronger terminals have a greater influx of Ca2+ per unit length during action potential invasion, but that in the absence of external Ca2+, tetanic stimulation results in comparable release of Ca2+ from internal stores in all terminals and comparable accumulation of Ca2+ in some large compartment, the subsequent emptying of which determines the time course of PTP.