To analyse evoked acetylcholine (ACh) release in the electric organ of Torpedo marmorata, a loose patch-clamp technique was used that allowed with a single extracellular electrode both focal depolarization of nerve endings and recording of the post-synaptic currents produced by the released transmitter. Two different types of post-synaptic response could be evoked by depolarizing pulses of increasing intensity: a graded response appearing with a delay of 0.6 ms (pulses of 0.2 ms duration), and an all-or-none response characterized by a mean delay of 1.4 ms. Both responses had a similar maximal amplitude and a similar rise time of 0.6 ms. The graded response was evoked in all places where spontaneous miniature electroplaque currents (m.e.e.s) could be recorded. It was not modified by 1 microM-tetrodotoxin (TTX), but was Ca2+ dependent and was abolished by Cd2+ (0.2 mM) or Mg2+ (10 mM). The all-or-none response could be evoked in only 30% of places where m.e.c.s. were recorded, it was highly TTX sensitive, Ca2+ dependent, and abolished by Cd2+ (0.2 mM) or Mg2+ (10 mM). K+ channel blocking agents, such as 4-aminopyridine (4-AP) or tetraethylammonium (TEA), which are known to prolong the duration of action potentials, prolonged the delay of the all-or-none response, but not that of the graded response. At low strength stimulation, the graded response was clearly evoked in a quantal way, with the quantum corresponding to the amplitude of spontaneous m.e.c.s. The amplitude distribution of the evoked responses closely followed a Poisson distribution. The maximum synchronous release of transmitter was found to be approximately 1.3 quanta/micron2 of presynaptic membrane and a mean quantal size of about 7000 ACh molecules was estimated from the charge transfer of m.e.c.s. The nerve terminal time constant was calculated from strength-duration curves obtained with depolarizing pulses just able to evoke either the all-or-none response or the first few quanta of the graded response. Respective mean values of 0.22 and 0.40 ms were found. Increasing the duration of the depolarizing pulse had two consequences: it differently affected the delay of the all-or-none response and that of the graded response; it increased the mean quantal content of the graded response. Both effects could not simply be accounted for by the influence of the nerve terminal time constant.(ABSTRACT TRUNCATED AT 400 WORDS)