1. Presynaptic receptor-mediated modulation of stimulation-evoked [3H]-acetylcholine[( 3H]-ACh) release from the neuromuscular junction was studied in the region of the mouse hemidiaphragm which contains the motor endplates, and which can easily be loaded with [3H]-choline. This method made it possible to detect exclusively the [Ca2+]0-dependent, quantal release of [3H]-ACh in response to axonal stimulation. 2. Atropine enhanced, and non-depolarizing muscle relaxants [+)-tubocurarine, pancuronium and pipecuronium) reduced, the release of [3H]-ACh evoked by high frequency trains of stimulation (50 Hz, 40 shocks) of the phrenic nerve. The effect of (+)-tubocurarine was frequency-dependent as at 5 Hz (40 shocks) it was less effective than at 50 Hz. The resting release of [3H]-ACh was not affected by these compounds. These findings indicate that ACh released into the synaptic gap by axonal firing reaches a concentration sufficient to influence its own release by a prejunctional effect. 3. The anticholinesterase, physostigmine sulphate, enhanced the release of [3H]-ACh in a concentration-dependent manner. This effect was mediated via prejunctional nicotinic receptor stimulation: (+)-tubocurarine, pancuronium and pipecuronium completely prevented the effect of physostigmine. 4. When the prejunctional nicotinic and muscarinic receptors were stimulated by a high concentration of extracellular ACh which had accumulated in the junctional gap in the presence of physostigmine, atropine did not influence the evoked release of [3H]-ACh. However, when the effect of endogenous ACh on nicotinic receptors was prevented by (+)-tubocurarine, atropine enhanced the release. 5. It is concluded that quantally-released ACh from motor endplates is subject to prejunctional automodulation: (a) ACh facilitates its own release via an effect on prejunctional nicotinic receptors (positive feedback), (b) ACh release is reduced by an action on muscarinic receptors. When the nicotinic receptor-mediated facilitation is fully operative, the muscarinic receptor-mediated negative feedback is much less effective. It is supposed that there is a link between the two feedback mechanisms possibly at the level of the second messenger system(s).