With the popularity of mobile networks and intelligent terminals, online fragmented learning, as a new learning method, has become the mainstream way for college students to acquire knowledge and study independently. However, college students are prone to “accept-interruption” in online fragmented learning; thus, it is difficult for them to master a complete knowledge system and form a rigorous logic system, which is essential to ensure the effect of online fragmented learning. Therefore, this study investigates the antecedents of college students’ continuance behaviors in online fragmented learning (CBOFL). Based on the expectation confirmation model (ECM), a theoretical model is developed to examine the factors influencing college students’ CBOFL. Taking a total of 429 undergraduate students who have studied contest courses on the Chinese university massive open online courses (MOOCs) for research subjects, the mechanism underlying the determinants of college students’ CBOFL is analyzed, and six hypotheses are tested by a structural equation modeling (SEM) technique with AMOS. The results indicate that confirmation positively impacts intrinsic learning motivation and satisfaction; intrinsic learning motivation, satisfaction, and teachers’ influence all significantly positively affect college students’ CBOFL. Additionally, the predicting powers of different factors on college students’ CBOFL vary broadly; therein, satisfaction has the most significant effect. This study makes theoretical contributions to the quantitative research on college students’ CBOFL and literature on the ECM. Still, it also has important practical significance in guiding college students’ CBOFL and facilitating the sustainability of online fragmented learning.