1. The cholinesterase inhibitors neostigmine, edrophonium and eserine (7 x 10(-7) M) reduced m.e.p.p. frequency by some 50% at the frog neuromuscular junction. Neostigmine also produced a small reduction in quantal content. 2. Tetraisopropylpyrophosphoramide, with a high specificity for non-specific cholinesterase, has a similar effect on m.e.p.p. frequency but ambenonium, with a high specificity for acetylcholinesterase, was markedly less effective in this respect. 3. Carbachol (10(-5) M) and the muscarinic agonists muscarine and metacholine (7 x 10(-7) M) also reduced the rate of spontaneous release. 4. The action of neostigmine was antagonized by atropine, but not by D-tubocurarine. Muscarine did not have any further effect when m.e.p.p. frequency was reduced with neostigmine. 5. Experiments with reduced extracellular Ca2+ concentration suggest that the cholinergic agents reduce Ca2+ permeability directly at the presynaptic terminals and that they do not act via a change in PNa or PK. It is suggested that the consequent reduction in Ca2+ entry causes a fall in both evoked release and in intracellular Ca2+ concentration, thereby reducing m.e.p.p. frequency. 6. It is concluded that non-specific cholinesterases present on the presynaptic terminals can act as inhibitory muscarinic cholinergic receptors. This form of presynaptic inhibition at the amphibian neuromuscular junction contrasts with that described in the mammalian preparation in which the sites are blocked by D-tubocurarine and are excitatory.