1. The actions of 0-1 mM ouabain and of K-free Ringer have been examined at the frog neuromuscular junction. 2. After a delay of more than 30 min, ouabain produces an increase in the miniature end-plate potential (m.e.p.p.) frequency. This increase occurs unchanged in Ca-free Ringer containing 1 mM-EGTA and is therefore unlikely to be due to an entry of Ca into the motor nerve terminals. 3. If the nerve to the preparation is stimulated repetitively in Ca-free Ringer containing 0-1 mM ouabain and 1 mM-EGTA the response of the m.e.p.p. frequency depends on the timing of the tetanus relative to the beginning of the ouabain treatment. 4. During the first 30 min of exposure to ouabain, the tetanus produces a small, transient increase in the m.e.p.p. frequency similar to that which occurs before ouabain is present. After about 30 min the same tetanus produces large, irreversible increases in the m.e.p.p. frequency. 5. Superfusion of an end-plate with K-free Ringer causes an immediate exponential rise in the m.e.p.p. frequency that is unaffected by the presence of external Ca ions. On replacing the normal K of the Ringer (2 mM) the m.e.p.p. frequency recovers quickly to its original value. 6. Late in an exposure to 0-1 mM ouabain the m.e.p.p. frequency becomes extremely sensitive to changes in the external Na concentration, [Na]o. Reducing [Na]o increases the m.e.p.p. frequency. The sensitivity to [Na]o is independent of external Ca ions or whether the isotonic substitute for NaCl is LiCl or sucrose. 7. It is suggested that the spontaneous release of transmitter is facilitated, in some way, by the changes in the monovalent cation content of the nerve terminals that result from blocking the Na-K exchange pump. The Na sensitivity of the m.e.p.p. frequency that develops simultaneously can be explained if a Na-dependent Ca efflux system is present in the membrane of the presynaptic terminals.