Four pigeons pecked keys for food reinforcers delivered by variable interval 1-min schedules during two successive 50-min sessions or one 100-min session. When 50-min sessions were conducted, they were separated by a 0-, 10- or 30-min delay, spent either inside or outside of the experimental enclosure. Responding usually increased to a peak and then decreased within sessions. This pattern was not altered by either the length of the delay between sessions or by the place where the subjects spent the delay. These results suggest that the beginning of the session or the passage of a short time between sessions restores the conditions necessary to produce within-session changes in responding. The results are incompatible with theories that explain these changes in terms of recovery from the handling routine, accumulation of arousal, priming, fatigue, or satiation. They can be reconciled with theories that attribute within-session changes in responding to reinstatement of memory or changes in attention only by arguing that memory and attentional adjustments are erased quickly between sessions.