Objective: To explore the longitudinal relationship between regulatory emotional self-efficacy and the persistence of exercise, so as to provide a reference for promoting the development of exercise habits among college students. Methods: Using a cluster random sampling method, one undergraduate institution was selected from each of Jiangxi, Hunan, Guizhou, Yunnan, Guangdong, Anhui, and Fujian provinces. A total of 2 576 college students were recruited. The Regulatory Emotional Self-efficacy Scale and Exercise Persistence Scale were distributed to college students through the questionnaire star platform. Three rounds of questionnaire surveys were completed in September 2023 (T1), December 2023 (T2), and March 2024 (T3). Crosslagged analysis was employed to explore the relationship between regulatory emotional selfefficacy and the persistence of college students exercise. Results: Repeated measures analysis of variance indicated that the main effects of regulatory emotional self-efficacy on time and gender were statistically significant ( F =102.15, η 2=0.07; F =34.80, η 2=0.01), and the interaction effects between time and gender, as well as between time and academic stage, were also statistically significant ( F =3.81, η 2=0.00; F = 25.54 , η 2=0.02)( P ＜ 0.05). The main effect of exercise persistence on time was statistically significant ( F =111.28, η 2=0.05) and the interaction effect between time and academic stage was statistically significant ( F =27.13, η 2=0.02)( P <0.01). Cross lagged analysis revealed that regulatory emotional self-efficacy at T1 positively predicted exercise persistence at T2 ( β =0.068), and regulatory emotional self-efficacy at T2 positively predicted exercise persistence at T3 ( β =0.368)( P ＜0.01). Prior exercise persistence positively predicted subsequent exercise persistence, with path coefficients of 0.298 and 0.240 ( P <0.01). Exercise persistence at T1 negatively predicted regulatory emotional self-efficacy at T2 ( β =-0.068), and exercise persistence at T2 positively predicted regulatory emotional self-efficacy at T3 ( β =0.061) ( P <0.01). Prior regulatory emotional self-efficacy positively predicted subsequent regulatory emotional self-efficacy, with path coefficients of 0.271 and 0.639 ( P <0.01). Conclusions There is a longitudinal causal relationship between regulatory emotional self-efficacy and college students exercise persistence. In daily physical activities, the significant role of emotional factors in sports practices should be emphasized to promote exercise behaviors among college students.