Rats received unsignaled shocks in an observation chamber, with different groups varying with respect to time between shocks. Twenty-four hours later the rats were returned to the observation chamber for a test of conditioning to contextual stimuli. The freezing response served as the dependent variable. In Experiment 1 we found that distributed shock trials (60 s) resulted in more context conditioning than did massed trials (3 s or 16 s). Experiment 2 replicated this intertrial interval (ITI) effect when total time in the context was equated for the massed and distributed groups. The observed beneficial effect of distributed practice for conditional stimuli arises because of decreased contextual conditioning with longer ITIs (e.g., Gibbon & Balsam, 1981; Rescorla & Wagner, 1972). Although the basic effect of enhanced performance with longer ITIs is consistent with Wagner's rehearsal model (e.g., 1978), three findings argue against such an account. First, posttrial stimulation did not reduce the benefit obtained from distributed trials (Experiment 3). Second, intertrial distractors did not improve performance of the animals subjected to massed trials (Experiment 4). And third, the ITI effect was not eliminated when conditioning was brought to its asymptote (Experiment 5). The overall pattern of data is consistent with an opponent-process account suggesting that in addition to supporting conditioning, the unconditional stimulus (US) activates a B-state capable of reducing the impact of the next US and that this B-state lasts longer than 16s but decays before 60 s.