Two experiments were conducted demonstrating that under certain conditions pigeons may peck at a higher rate on a key that produces intermittent reinforcement following a delay than on one that always produces reinforcement following the same delay duration. In both experiments, concurrent chain schedules were employed. In Experiment I, a single peck on one key led to a white light and a delay of 15 sec, which always terminated with food. A peck on the other key led to its illumination by one of two colored lights and a delay period of 15 sec. The delay was followed by either food presentation or timeout, either one lasting 3 sec. In a control group, the lights on this key were not correlated with food or timeout. Under the correlated stimuli, birds more often pecked the key leading to intermittent reinforcement, whereas with uncorrelated stimuli they pecked the key leading to the white light and 100% reinforcement. In Experiment II, concurrent variable-interval schedules were employed in the first link. The results showed generally that the relative rate was higher on the key leading to intermittent reinforceimient when the stimuli were correlated with reinforcement and timeout than on the key leading to 100% reinforcement. There was some indication that this performance was affected by (1) the duration of the delay, (2) the percentage of reinforcement on the key yielding the higher percentage of reinforcement (the key with the white light), and (3) prior experimental conditions. Using a modified chain schedule procedure, Wilton and Clements (197 la) correlated with only one of the two stimuli in the terminal link; the other stimulus was followed by a short timeout. In another condition, food was delivered at the end of half of the terminal links, but neither stimulus was reliably correlated with its delivery. In the third condition, each terminal link, regardless of the stimulus, terminated with food delivery. The response rate in the initial component was highest in the first of these conditions, without much difference between the last two conditions. A study by Kendall (1972) replicated these findings with terminal links of two different durations and reinforcement in each. Response rates were higher in the initial component followed by one long and one short terminal component than in the component followed by two short terminal components. This effect disappeared, however, when the two stimuli for the terminal components were no longer reliably correlated with the duration of the terminal components. Wilton and Clements (1971a) and Wilton (1972) interpreted the findings of the Wilton and Clements experiments in terms of an information hypothesis.