As a target memory is recalled or practiced, its competitors are suppressed, making them more difficult to recall in future attempts. This is called retrieval-induced forgetting (RIF; Anderson, Björk, & Björk, 1994), and is evidenced by lesser recall for unpracticed members of a partially practiced list (RP- items) than for members of an unpracticed list (NRP items). The RIF effect, while extremely robust, is not immutable. Directed integration of items into a unified knowledge structure has been shown to greatly reduce or even eliminate the RIF effect (Anderson & McCulloch, 1999). Additionally, high associate word lists, those with greater semantic relatedness, have been shown to result in spontaneous integration (Bäuml & Kuhbandner, 2003). In this study, we combined these factors, along with delay, into a 3 x 2 x 2 x 2 (integration x associativity x delay x set) mixed-model design, using Deese-Roediger-McDermott (DRM) word lists (Stadler, Roediger, & McDermott, 1999), in order to examine the operation of these mediating factors. While set (RIF effect) and delay were statistically significant, we observed no main effects of either integration or associativity. Additionally, there were no statistically significant interactions, indicating that RIF can be robust even to those factors, like integration and associativity, which under certain circumstances reduce its effects.