Multiple-choice questions (MCQs) are popular in vocabulary software because they can be scored automatically and are compatible with many input devices (e.g., touchscreens). Answering MCQs is beneficial for learning, especially when learners retrieve knowledge from memory to evaluate plausible answer alternatives. However, such retrieval may not always occur (e.g., with easy-to-guess answers). Therefore, we tested whether we could optimize MCQs for retrieval practice with a "stepwise display," which presents the question before the answer options. This creates an opportunity for cued recall. In an experimental classroom study (N = 75) and three online experiments with adult participants (N = 45, N = 77, and N = 79), participants practiced vocabulary with either standard MCQs or stepwise MCQs: In standard MCQs, a word was presented with three possible translations; in stepwise MCQs, it was shown for 4 s before the translations appeared. In three of the four experiments (Experiments 1, 3a, and 3b), stepwise MCQs enhanced retention significantly compared to standard MCQs, measured by a posttest several days after practice. Benefits of stepwise MCQs were found for different translation directions (Experiments 1 and 3) and both with easy-to-guess answers (noncompetitive answer options, Experiment 1) and hard-to-guess answers (competitive answer options, Experiment 3). However, the effect was most robust for questions with noncompetitive answer options (Experiments 1, 3a, and 3b) and for learners who reported trying to retrieve the answer from memory during the stepwise display (Experiments 1 and 3). Moreover, retention was generally enhanced by competitive, hard-to-guess answer options (Experiments 3a and 3b). Overall, a stepwise display is a promising and easy-to-implement manipulation to optimize MCQs for retrieval practice.