We investigated the effect of South American coral snake (Micrurus lemniscatus lemniscatus) venom on neurotransmission in vertebrate nerve-muscle preparations in vitro. The venom showed calcium-dependent PLA2 activity and caused irreversible neuromuscular blockade (0.1-30 µg/ml) in chick biventer cervicis (BC) and mouse phrenic nerve-diaphragm (PND) preparations. In BC preparations, contractures to exogenous acetylcholine and carbachol (CCh), but not KCl, were abolished by venom concentrations ≥0.3 µg/ml; in PND preparations, the amplitude of the tetanic response was progressively attenuated, but with little tetanic fade. In low Ca 2+ physiological solution, venom (10 µg/ml) caused neuromuscular blockade in PND preparations within ~10 min that was reversible by washing; the addition of Ca 2+ immediately after the blockade temporarily restored the twitch responses, but did not prevent the progression to irreversible blockade. Venom (10 µg/ml) did not depolarize diaphragm muscle, prevent depolarization by CCh, or cause muscle contracture or histological damage. Venom (3 µg/ml) had a biphasic effect on the frequency of miniature end-plate potentials, but did not affect their amplitude; there was a progressive decrease in the amplitude of evoked end-plate potentials. The amplitude of compound action potentials in mouse sciatic nerve was unaffected by venom (10 µg/ml). Pre-incubation of venom with coral snake antivenom (Instituto Butantan) at the recommended antivenom:venom ratio did not neutralize the neuromuscular blockade in PND preparations, but total neutralization was achieved with a 10-fold greater volume of antivenom. The addition of antivenom after 50% and 80% blockade restored the twitch responses. These results show that M. l. lemniscatus venom causes potent, irreversible neuromuscular blockade, without myonecrosis. This blockade is apparently mediated by pre-and postsynaptic neurotoxins and can be reversed by coralsnake antivenom.