1. The hypothesis is put forward that a residue of the ;active calcium' which enters the terminal axon membrane during the nerve impulse is responsible for short-term facilitation.2. This suggestion has been tested on the myoneural junction by varying the local calcium concentration so that during the first of two nerve impulses [Ca](o) is either much lower than, or raised to a level approaching that, during the second impulse. Facilitation is much larger in the latter case, which is in accordance with the ;calcium hypothesis'.3. A short pulse of depolarization focally applied to the junction is followed by a brief period of very intense facilitation. This can be seen in the tetrodotoxin-treated preparation, e.g. by lengthening the depolarization from 1 to 2 msec which can cause a more than fifty-fold increase in transmitter release. This large ;early facilitation' (which presumably occurs also during the course of a normal action potential) is discussed in relation to the ;calcium hypothesis'.