In a previous directed-forgetting experiment using a paired-associate probe technique, occasional paired associates which 11 female and 7 male undergraduates had been instructed to forget were repeated in the set which was to be remembered. The results suggested that subjects were aware of that repetition and concentrated their rehearsal capacity on those items, to the detriment of non-repeated ones. The present experiment was an analogue of the earlier one but utilized CCC trigrams rather than paired associates, and the data were scored as for free recall of three unrelated letters. It was reasoned that the repetition of a letter from one CCC to the next would not be recognized by subjects as repetitions. The results showed that, when a letter in a trigram the subject had been instructed to forget was repeated, it was recalled no better than were non-repeated letters. When a letter was repeated in a condition in which there was no instruction to forget, however, recall of the repeated letter improved. It was argued that in this experiment, the instruction to forget caused the repeated letter to be functionally differentiated in two trigrams, and earlier rehearsal did not benefit its retention. When there was no instruction to forget, the repeated letter was functionally identical in both trigrams, and the extra rehearsal improved its level of recall.