Studies of weak-light onset reinforcement in rats have used lights of fixed durations ranging from M to 5 sec., or have left the light on as long as the lever was held down. The present experiment tested the effect of the light's duration upon conditioned lever pressing. The apparatus was a Skinner box made from a rat living cage and containing a 1-in.-wide lever. A 25-w frosted lamp bulb was mounted outside the cage, 5 in. above the bar. A resistor reduced the lamp voltage to 95 v. S was first left in the darkened box for a 10-min. habituation period. Then, 10 light-onset reinforcements were given and were followed immediately by a 20-min. extinction period in darkness.' For one group of 16 Ss (female yearold abino rats), the light stayed on for y3 sec., regardless of the duration of the press. Other groups of 16 received lights lasting 1, 3, and 9 sec. Effect of the conditioning was measured by the number of presses during extinction. For the Y3-, I-, 3-, and 9-sec. groups, the mean frequencies were 11.4, 14.2, 15.8, and 13.1 presses, respectively. Simple analysis of variance did not reveal a duration effect approaching significance. In fact, the withingroups mean square exceeded that between groups, although not by a significant a m o ~ n t . ~ These results suggest that reinforcement results from the change from darkness to light, rather than from the light itself. Thus they lend some support to the stimulus-change interpretation of weak-light reinforcement. Other evidence, however, such as the usual failure to obtain reinforcement by light offset (e.g., Robinson, 1959) seems inconsistent with the hypothesis that all stimulus changes are reinforcing to rats.