A small endplate potential can be recorded in frog muscle fibres, after all acetylcholine-mediated transmission has been eliminated by pre- or postsynaptic blocking agents (botulinum toxin, calcium lack, manganese, curare, alpha-bungarotoxin). It is usually necessary to hyperpolarize the muscle membrane to detect this 'non-cholinergic' endplate potential. Below--100 mV little or no response is seen; a maximum is reached at about--140 mV, when the amplitude can be as large as 100 microV (endplate current up to about 1 nA). Other characteristic features are: the response shows no quantal fluctuations; its amplitude is not facilitated by repetitive impulses; its size and time course are not noticeably affected by prostigmine, curare or alpha-bungarotoxin; the half-time of decline of the endplate current is approximately 1.7 ms at 20 degrees C, and is lengthened by lowering the temperature with a Q10 of about 1.3; the response is abolished by barium. When iontophoretic pulses of potassium are applied to the endplate, local depolarization is recorded whose amplitude varies with membrane potential similarly to that of the nerve-evoked response. These observations strongly indicate that this 'non-cholinergic', 'non-quantal' endplate potential arises from a rapid synaptic transfer of potassium ions, released by the active nerve terminal into the synaptic cleft and entering the muscle fibre through 'anomalous rectifier' channels in the endplate membrane.