Rats were trained in a two-way shock-avoidance task, either in a single training session of 100 trials or over a 5 day period with 20 trials per day. Electroencephalograph and electromyograph recordings were continuous for 2 days prior (base line), during, and following the experiments, except during the actual training sessions. Animals that learned the task showed marked increases in PS. Rats that learned the distributed trials training task showed PS increases in the 24 hr period prior to a maximum increase in correct performance. These PS elevations appeared after a 10 hr latency following the training session. The animals that learned the single training session exhibited PS increases that persisted for a period of 6 days, which appeared daily in a cyclic manner. The latency to onset of these PS changes was less than 1 hr. The PS increases in both types of training were due to an increase in the number of PS periods and are believed to be different manifestations of the same flexible mechanism. This mechanism is considered to be involved with some aspect of learning other than the acquisition process, such as retention.