It has been reported, previously, that as escape-from-shock training progressed, Ss increasingly held the lever down. This was true both when shock termination was contingent upon a lever press (press training) and when it was contingent upon release following a press (release training). Moreover, changes in holding were not primarily reflected by changes in response duration but, rather, by shifts in the relative frequencies of responses which were held until the reintroduction of shock, particularly, in press training, escapes which were held until the reintroduction of shock (punished escapes) and, in release training, extra responses which were held until shock returned (punished extras). Finally, the writer had observed some indications that interruption of release training temporarily reduced holding. The present study was undertaken to determine whether the decremental effect of rest on holding is a reliable phenomenon and whether it is obtained with press as well as with release training. Ss were trained at one of four degrees of inter-session time (0, 1, 5, or 10 days of rest) and under either the press or release training procedure. In press training the over-all effect of inter-session time on holding (change in relative frequency of punished escapes) was not significant. The pattern of effects was comparable to those attributed to fatigue or other inhibitory factors which dissipate rapidly during rest. In release training, the effect of inter-session time on holding (change in relative frequency of punished extras) was predominantly decremental. That is, in release training, rest reduced holding. These results are comparable to those commonly attributed to the establishment of competing or interfering stimulus-response relations during rest. A theoretical analysis of response chaining accounted for the interaction of training procedure and inter-session time and indicated, as did supporting data, that delays in training reduced the strength of extra presses, i.e., their initiation, and did not reduce holding per se.