Recall direction is known to be an important determinant of serial recall performance: For example, accuracy is often greater for forward recall than backward recall, and forward recall typically exhibits extensive primacy but little recency, with the reverse arrangement for backward recall. Although some of the differences between recall directions can be accommodated by models that postulate a single retrieval process, recent evidence appears to favor the existence of 2 distinct retrieval processes, 1 for forward and 1 for backward recall. Five experiments reported in this article were aimed at illuminating these 2 putative processes. Tasks that interfered with the formation of interitem associations at study were found to disrupt forward but not backward recall, whereas tasks that altered the visual-spatial characteristics of the study material affected backward but not forward recall. It was proposed that forward recall is largely based on interitem associa tions, whereas backward recall relies on a visual-spatial representation of the study material.