This study investigates the impact of task-induced involvement and time on task on incidental second language (L2) vocabulary acquisition. Utilizing a 3 (task-induced involvement) × 2 (time on task) × 2 (post-test time) research design, three task-induced involvement conditions were employed based on the Involvement Load Hypothesis (ILH): reading and gap-fill task, reading and sentence-making task, and reading and translation task, with corresponding involvement load (IL) indices of 2, 3, and 4, respectively. Two time-on-task conditions were implemented: uncontrolled time on task, where participants in different groups completed tasks with varied durations, and controlled time on task, where participants in different groups completed tasks with roughly equal durations. Five intact classes comprising 256 Chinese middle school students participated and were randomly assigned to one of five designed tasks aimed at learning 10 carefully selected target words. The results of a three-way repeated measures ANOVA indicate a significant three-way interaction effect among task-induced involvement, time on task, and post-test time, as well as a significant two-way interaction effect between task-induced involvement and time on task. These findings demonstrate that task-induced involvement and time on task interact to significantly influence both initial acquisition and retention of incidental L2 vocabulary. Specifically, under uncontrolled time conditions, tasks with higher ILs and longer durations yield better initial vocabulary gains and retention, partially supporting the ILH. Conversely, under controlled time conditions, tasks with lower ILs exhibit superior initial vocabulary gains and retention, contradicting the predictions of the ILH. Relevant implications are also discussed.