Memory retrieval often enhances later memory compared with restudying (i.e., the testing effect), indicating that retrieval does not simply reveal but also modifies memory representations. Dividing attention (DA) during encoding greatly disrupts later memory performance while DA during retrieval typically has modest effects--but what of the memory-modifying effects of retrieval? If these effects are similar to study-based encoding, they should be greatly disrupted by DA, a possibility consistent with elaborative and effortful accounts of the testing effect. Alternatively, the mnemonic consequences of retrieval may be largely resilient to distraction, like retrieval itself. In 3 experiments, participants studied word pairs (Phase 1) then engaged in restudy of some pairs and retrieval of others (Phase 2), followed by a final cued-recall test (Phase 3). Phase 2 restudy and retrieval occurred under full attention (FA) or DA. The experiments were designed to induce either material-specific (Experiments 1 and 2) or material-general (Experiment 3) interference, as well as to produce comparable secondary task performance between the restudy and retrieval groups (Experiments 2 and 3). Consistent with prior research, retrieval improved final recall (i.e., the testing effect) whereas DA disrupted final recall. Critically, the 2 factors interacted such that the negative effect of DA on final recall was substantial in the restudy condition but quite modest in the retrieval condition--resulting in a larger testing effect in the DA than FA condition. The encoding effects of retrieval seem resilient to distraction which has implications for theories of the testing effect.