Cumulative tests have been shown to enhance second language (L2) vocabulary learning by incrementally incorporating both new and previously learned materials in daily or weekly quizzes. In this study, we investigated the role of three factors that could impact the effectiveness of cumulative testing: (a) study time and (b) spaced practice (review counts) to prepare for the test, and (c) L2 proficiency. We assigned Japanese university students from two intact classes to the cumulative (n = 44) or noncumulative (n = 44) group. They studied 100 words for weekly quizzes over a semester, with 10 items introduced in each class. The cumulative group was tested simultaneously on newly and previously studied words, while the noncumulative group was tested on the most recent set of 10 items. Students prepared for quizzes using online flashcards tracking the number of practice trials and review counts. Posttest results showed that the cumulative group outperformed the noncumulative group, but this advantage depended on the amount of practice and L2 proficiency. Specifically, the cumulative group showed superiority only with an increasing amount of practice, especially among those students with lower L2 proficiency. Surprisingly, the two groups did not differ in the overall study time, indicating that students in the cumulative group did not increase study time but strategically chose which previously studied words to review. Our findings demonstrated not only whether but why cumulative tests work better than noncumulative tests.