||Spaced practice generally leads to better long-term recall than massed practice. The present study examined the conditions under which college students choose to mass or space practice when given the choice. Students studied word pairs of varying difficulty and decided whether they wanted to study each pair again, and, if so, whether to study it immediately (massed practice) or later (spaced practice). The likelihood of choosing spaced practice did not vary with a priori item difficulty but was higher when pairs were presented for 5 secs rather than 1 sec. The former result is most consistent with the Discrepancy Reduction model, as interpreted by Benjamin and Bird (2006), whereas the latter result is more consistent with the Region of Proximal Learning model (Metcalfe & Kornell, 2005). In the discussion, we consider how these findings, and findings from previous work (i.e., Son, 2004), might be reconciled within a single explanatory framework.