1. Experiments were done on Mg-blocked neuromuscular junctions of the frog to study in detail the magnitude and time course of facilitation of the end-plate potential (e.p.p.).2. When two shocks were applied to the motor nerve with an interval of 5 msec or less between them, the amplitude of the second e.p.p. was facilitated by approximately 100%. This facilitation decreased as the interval was lengthened.3. The facilitation could be separated into two components on the basis of the time course of its decay as the interval between the two shocks was increased. The first component decayed exponentially after the conditioning shock with a time constant of about 35 msec. The second became apparent 60-80 msec after the conditioning shock, rose to a maximum of approximately 15% at about 120 msec and decayed thereafter with a time constant of the order of 250 msec.4. The growth of e.p.p. amplitude during short trains of repetitive stimulation at frequencies up to 100/sec and its subsequent decay could be predicted by assuming that the magnitude and time course of both components of facilitation were the same for every shock in the train and that these individual facilitatory effects summed linearly.