1. End‐plate potentials (e.p.p.s) and miniature end‐plate potentials (min.e.p.p.s) were recorded intracellularly from the cutaneous pectoris nerve‐muscle preparation of the frog during prolonged stimulation at low frequencies (5/sec—50/sec). 2. When Ca was present in the bathing solution, the quantum content of the e.p.p. and the frequency of occurrence of the min.e.p.p.s gradually increased during the period of stimulation. During the first few minutes of stimulation, the min.e.p.p. frequency increased linearly with time, and the rate of increase was dependent on the Ca concentration of the bathing solution. However, Mg had no effect on this Ca‐dependent increase in min.e.p.p. frequency. 3. A large maintained increase in min.e.p.p. frequency also occurred during prolonged stimulation in solutions that contained no added Ca and 1‐2 m M‐EGTA. Under these conditions the increase in min.e.p.p. frequency was dependent on the Mg concentration of the bathing solution and was exponential in time. 4. It is suggested that the rise in min.e.p.p. frequency is caused by an accumulation of Ca or Mg ions in the nerve terminal, and it is suggested that these ions enter the terminal at relatively non‐specific sites distinct from the Ca‐specific sites that trigger the ‘phasic’ release of transmitter.