The extent to which a stimulus established as an effective cue to forget using one set of sample stimuli would function also as an effective cue to forget using a second set of sample stimuli was investigated. In the first two experiments, an effective cue to forget was occasionally presented on trials involving sample stimuli which had not previously been followed by any type of postsample stimulus. Neither experiment revealed any evidence of transfer of cue effectiveness across samples from different dimensions. In the final experiment, cues to remember and to forget were established initially on trials involving samples of food and no food and horizontal and vertical lines. One of the two cues was then presented on each trial involving color sample stimuli, and the contingencies were such that one of the cues, a triangle, would be expected to function as a cue to remember and the other, a circle, would be expected to function as a cue to forget. The cuing function of the circle and triangle was the same on all three matching problems for some birds, whereas the cuing function of the circle and triangle differed across problems for other birds. Accuracy on cued red/green trials was influenced strongly by whether the cues had consistent or inconsistent significance across the three matching tasks. The data were interpreted within a perspective in which a distinction is maintained between learning how to terminate rehearsal and learning when to terminate rehearsal.