The word length effect has been a central feature of theorising about immediate memory. The notion that short-term memory traces rapidly decay unless refreshed by rehearsal is based primarily upon the finding that serial recall for short words is better than that for long words. The decay account of the word length effect has come under pressure in recent times. The current research tests alternative explanations of the word length effect, ones that suggest first, that word length effects should be found in long-term memory tasks. Second, one alternative predicts that the standard short-word advantage should only be observed in tasks in which participants use serial order memory. In tasks where only item memory is used, the approach predicts a long word advantage. Consistent with this notion, standard word length effects were found in long-term serial recall and free recall tasks, but a long word advantage was found in item recognition and cued recall tasks.