Pigeons were tested in an intratrial interference preparation in which two stimuli (red and green fields) were presented successively as samples and then simultaneously as comparison stimuli. Choice of the comparison matching the second sample was designated correct and was reinforced. On different trials, either the first or second sample was followed by a cue to forget (horizontal line), and the alternate sample was followed by a cue to remember (vertical line). Relative to baseline trials in which both samples were followed by a cue to remember, accuracy was enhanced on trials in which the forget cue was presented immediately following the first sample and was reduced on trials in which the forget cue was presented immediately following the second sample. These findings suggest that presentation of a forget cue can act to reduce memory differentially for one of two samples presented on a trial. It was concluded that the context-retrieval interpretation of directed forgetting could not account for this pattern of findings. It was concluded further that the rehearsal interpretation of directed forgetting provided a ready account of the data, an account which was tested and supported in the final experiment.