1. Transmitter release at crayfish and frog neuromuscular junctions was studied by recording synaptic potentials with extracellular or intracellular glass micro-electrodes. 2. The binomial release parameters n and p were calculated using the experimental observations for the mean number of quanta (m) released in a series of trials, the variance (var) of the quantum content distribution and the number of transmission failures (n-o). 3. In one series of experiments on frog neuromuscular junction, action potentials were blocked by tetrodotoxin. Transmitter release was limited to a circumscribed part of the nerve terminal by focal stimulation with a glass micro-electrode. Values for m were between 0-6 and 5-7 and calculated values for n ranged from 3 to 23. p was between 0-05 and 0-48. In a second series of experiments on frog neuromuscular junction transmitter release was evoked by conventional stimulation of the nerve trunk in Mg-Ringer. m was mostly between 20 and 60 and in the same experiments n was calculated to be between 70 and 125. Values for p ranged from 0-25 to 0-48. 4. It can be concluded from the present results that transmitter release at frog neuromuscular junction is described by binomial statistics as is the case for neuromuscular junction of crayfish. 5. In comparing the values for n for the two series of experiments on frog neuromuscular junction it appears that n is dependent on the length of the activated synaptic contact. The numbers for n calculated from frog neuromuscular junction are of the same order of magnitude as the number of 'active zones' seen in the corresponding length of the synaptic nerve terminal.