The testing effect is 1 of several memory effects moderated by experimental design, such that the effect on free recall is larger in a mixed-list than pure-list design (Mulligan, Susser, & Smith, 2016). The current experiments assess hypotheses regarding why this pattern is found. Three extant accounts of design effects (Nguyen & McDaniel, 2015) are the item-order account, the retrieval-distinctiveness account, and the rehearsal-borrowing account. Three experiments contrasted these accounts, finding support for rehearsal borrowing but no evidence for the rehearsal-distinctiveness or item-order accounts. In addition, each experiment found that the testing effect was larger in the mixed-list than pure-list condition, attesting to the replicability of the design-moderation pattern for the testing effect.