In this study, potential benefits of early memory testing were examined in terms of "inoculating" eyewitness memory against forgetting. As predicted by fuzzy trace theory (e.g., Reyna & Titcomb, 1997), a larger testing advantage in the delayed recall of event details was expected after immediate testing than after delayed testing because of the decline in accessibility of verbatim traces over time. However, memory for only the gist of these details was expected to be relatively stable over time, resulting in a smaller (if any) effect of the timing of interpolated testing. After viewing a target event, participants were questioned about event items immediately, after a 24-hr delay, or after a 48-hr delay and were free to respond at either the gist or the verbatim level. Verbatim memory for event details was tested 72 hr after the event. As expected, immediate interpolated testing improved verbatim memory performance on the final test more than delayed testing did, yielding a larger testing effect. Furthermore, the effect of the timing of interpolated testing on the magnitude of the testing effect was mediated by verbatim accessibility at interpolated testing. In contrast, memory for only the gist of event details was unaffected by the timing of interpolated testing, both on the interpolated test and on the final test. The findings highlight the role of declining verbatim memory over time in accounting for the advantage of immediate over delayed interpolated testing in inoculating eyewitness memory against forgetting of detailed information. (Contains 3 figures and 6 footnotes.)