1. Mass fragmentography was used to measure the tissue content and release of acetylcholine (ACh) by frog sartorius muscles, which had been previously treated with an irreversible cholinesterase inhibitor. The frequency of miniature end-plate currents (m.e.p.c.s) was also measured. 2. Exposure of muscles for 15 min to 2 mM-LaCl3 resulted in a large release of ACh which subsided to low levels after 1 h. About 4 h later treatment with 50 mM-KCl, or with the calcium ionophore A 23187, or with a second dose of LaCl3, all failed to augment ACh release, notwithstanding the fact that the ACh content of La3+-treated muscles was about the same as that of controls. 3. Hypertonic NaCl or raised KCl concentrations were used to increase m.e.p.c.s and this also increased ACh release; it was estimated that each quantum corresponded to the release of 12 000 molecules of ACh. 4. ACh release by nerve stimulation was greatly potentiated by 10 mM-tetraethylammonium chloride, and this enabled the ACh released by ten, and even single, stimuli to be detected; it was calculated from the ACh released and the quantal content that each quantum contained on the average 13 000 molecules. 5. ACh released by nerve stimulation at 0.2/s in the absence of tetraethylammonium was about half that expected on the basis of previous estimates of quantal content; it was increased about two-fold by alpha-bungarotoxin. 6. It is concluded that chemical and electrical stimulation of the nerve evoked quantal ACh release, without influencing non-quantal ACh leakage. The results are consistent with the view that ACh quanta are derived from synaptic vesicles. They also show that resting ACh release is not due to leakage of ACh ions along an electrochemical gradient in the membrane.