What is an effective strategy to ensure high participation and persistent waste separation behavior? To address this question, we examined the effectiveness of monitoring and voluntariness in a model of the dynamic changes of actual behavior. Here, we conducted a large-scale longitudinal field study with four periods (baseline, the first intervention vs. withdrawal vs. the second intervention) for two strategies (monitoring vs. voluntariness) over a one-year period. One hundred and one students (Mage = 12.29 ± 0.58 years) in a junior high school took part in the field study. Our findings demonstrate that regular monitoring is necessary to encourage long-term waste separation behavior. Once the monitoring was removed or the voluntary policy was implemented, the weight of unsorted waste increased sharply and individual behaviors returned to that in the baseline period. Even when monitoring was readopted during the second intervention period, more efforts were needed for administrators to promote waste separation behaviors. Based on the social learning theory, we propose that regular monitoring plays a vital role in promoting long-term waste sorting behavior through feedback, observational learning, and habit-forming. The current study sheds new light on the critical role of long-term monitoring for successful waste separation management. Our work opens avenues for policymakers and has broad policy applicability from schools to communities.