The present experiments investigated the idea that the anticipated delay to initial recall governs whether a subject will rely on elaborative or maintenance rehearsal. In agreement with earlier research, the results suggested that maintenance rehearsal is used only when words can be rehearsed up to the moment of recall. A novel finding concerned the accessibility of the items processed in that way. When subjects were explicitly asked to recall those items first in final free recall, they were as likely to be recalled as items that had been processed elaboratively. This finding suggests that even rote rehearsal creates an accessible memory trace, albeit one that is most efficiently accessed by contextual rather than semantic cues.Following the introduction of the levels of processing framework, one prominent line of research investigated the effects of various orienting tasks on delayed memory performance (Craik & Lockhart, 1972;Craik & Tulving, 1975). The typical finding from this literature was that tasks involving semantic analysis lead to better recall of words than tasks involving structural or phonemic analyses (Lewandowsky & Hockley, 1987). A second line of research fostered by the levels of processing framework investigated the conditions under which subjects will naturally rely on different levels of processing when they are attempting to memorize a list of words. The central question addressed in this literature is whether subjects choose to devote semantic (or elaborative) rehearsal to some words and simple rote (or maintenance) rehearsal to others. Indeed, as described below, dynamic shifts in rehearsal strategy appear to be the norm. The experiments described herein were intended to further advance this line of inquiry, with particular emphasis on the conditions giving rise to and the enduring consequences of maintenance rehearsal.Most of the research on the issue of rehearsal strategy is consistent with the idea that subjects use some form of elaborative rehearsal unless it is clear that the items being rehearsed can be maintained indefinitely (i.e., up to the moment of recall). When that condition obtains, elaborative strategies are essentially superfluous and may be abandoned in favor of simple maintenance rehearsal (Gotz & Jacoby, 1974;Jacoby & Bartz, 1972;Meunier, Ritz, & Meunier, 1972;Roenker, 1974;Watkins & Watkins, 1974). In a standard list memory procedure, the initial items of a list cannot be rehearsed indefinitely because of the incoming stream of subsequent items. Thus, the initial items would seem to be the most likely candidates for elaborative rehearsal. The last few items of a list, on the other hand, can be rehearsed unencumbered until the recall period begins just a few moments later. As a result, these items would seem to be the most likely candidates for I would like to express thanks to Heidi Squire, Julie Baroway, Paul Havig, and Barbara Schneider for their assistance in data collection, and Robert Greene, Mark McDaniel, and Vito Modigliani for their thoughtful commentaries on ...