Two short-term memory experiments investigated the effects of instructions to forget. In Exp. I, proactive interference (PI) was reduced when 5s were cued to forget the first of two trigrams; but there was no reduction in retroactive interference when 5s were cued to forget the second. The residual amount of PI depended upon the acoustic similarity of the two trigrams. In Exp. II, PI was reduced when 5s were cued to forget the first 6 words in a 12-word sequence. A subsequent recognition test, however, showed that the to-beforgotten words were recognized as often as words taken from sequences with no forget cue. The results supported a hypothesis that the effects of a forget cue are due to differential storage and retrieval of to-be-remembered and tobe-forgotten items. Forgetting is usually considered to be an inevitable, undesirable result of the process