1. Frog sartorius muscles were incubated in Ringer solutions with a raised K+ concentration (high K+) and then allowed to recover in medium with a normal K+ concentration. During the recovery period miniature end-plate potentials (m.e.p.p.s) were recorded with intracellular electrodes. In addition, the acetylcholine (ACh) released from muscles in the presence of high K+ was measured by a mass spectrometric method. 2. Incubation in a high-K+ medium induced the appearance of giant miniature end-plate potentials (g.m.e.p.p.s). However, if the Cl- of the medium was substituted by propionate, very few g.m.e.p.p.s were observed. This was due to the absence of Cl- and not to the presence of propionate. The frequency of g.m.e.p.p.s was also greatly depressed when the Cl- concentration was lowered from 120 to 60 mM. 3. The amount of ACh released into a high-K+ medium was the same, regardless of whether Cl- or propionate was the anion. 4. When Cl- was replaced by NO3- or Br-, incubation in high-K+ Ringer solution induced the appearance of g.m.e.p.p.s. However, SO4(2-), like propionate, was unable to substitute for Cl- in this respect. 5. The frequency of g.m.e.p.p.s was correlated with that of m.e.p.p.s during the recovery period. However, when the K+ concentration was raised to 17 mM the frequency of m.e.p.p.s greatly increased, whereas that of the g.m.e.p.p.s did not change significantly. 6. G.m.e.p.p.s disappeared in the presence of curare, but persisted in the presence of tetrodotoxin or in a Ca2+-lacking medium. However, g.m.e.p.p.s failed to appear when the medium had lacked Ca2+ during the stimulation. 7. It is tentatively concluded that g.m.e.p.p.s are associated with Cl--dependent processes which occur after stimulation of transmitter release, and which are linked with the endocytotic retrieval of presynaptic membrane.