Taking a test improves memory for that tested information, a finding referred to as the testing effect. Multiple-choice tests tend to produce smaller testing effects than do cued-recall tests, and this result is largely attributed to the different processing that the two formats are assumed to induce. Specifically, it is generally assumed that the multiple-choice format bypasses the need to retrieve information. Research suggests, however, that multiple-choice questions can be constructed to induce retrieval of information pertaining to the incorrect alternatives. In the present research, we investigated the processes that individuals use to answer multiple-choice questions and how those processes relate to later memory, particularly for information pertaining to the incorrect alternatives. Most critically, we found that participants sometimes spontaneously recall information pertaining to incorrect alternatives, and these spontaneous retrievals are associated with retention of those alternatives as correct answers to related questions later. Although multiple-choice questions can be constructed so as to bypass retrieval, they can also be constructed to induce retrieval, and when they are, learning benefits are likely to occur. The present work has practical implications for how instructors can create multiple-choice questions to induce processes that facilitate learning.