1. Electrical activity of neuromuscular junctions of the frog was studied in a medium (Ca-Ringer) whose Na ions had been entirely replaced by Ca.2. Spontaneous miniature end-plate potentials (m.e.p.p.s) of reduced amplitude are recorded in this abnormal ionic environment, and graded end-plate potentials can be elicited by applying depolarizing current pulses to the pre-junctional parts of the nerve.3. Addition of 5 mM tetraethylammonium (TEA) to the Ca-Ringer causes the appearance, in almost all-or-none fashion, of very large e.p.p.s (up to 45 mV in amplitude) in response to nerve stimulation.4. These ;giant' e.p.p.s occur despite the curarizing action of TEA (and its depressing effect on the amplitude of m.e.p.p.s) and they persist after application of tetrodotoxin.5. After several hours exposure to Ca-Ringer, spontaneous end-plate activity gradually declines, and eventually evoked e.p.p. responses fail. On return to normal Na-Ringer, spontaneous end-plate activity is quickly resumed, but the potentials have an abnormal, very wide, amplitude distribution.6. The results are discussed, in conjunction with relevant work on the squid giant synapse, in terms of the ;calcium hypothesis' of transmitter release.