1. Depression of transmitter release produced by preceding conditioning stimulation was studied at the frog's neuromuscular junction.2. Depression occurred when transmitter release was restricted to a short length of nerve terminal, or when release was initiated by electrotonic depolarization of the terminals after action potentials were abolished by tetrodotoxin.3. Quantitative studies revealed a non-linear relationship between the estimated magnitude of ;zero-time' depression and the amount of transmitter released by conditioning stimulation.4. Two stimuli separated by 20-200 msec were given and the ratio of the end-plate potential amplitudes (V(2)/V(1)) was measured. This amplitude ratio increased during depression produced by stimuli preceding the test pair.5. These observations may be explained by assuming that depression is associated with a reduction in release probability as well as a depletion of transmitter available for release.