Background: Although extensive research has examined epistemological beliefs, reflective thinking and learning approaches, very few studies have looked at these three theoretical frameworks in their totality. Aims: This research tested two separate structural models of epistemological beliefs, learning approaches, reflective thinking and academic performance among tertiary students over a period of 12 months. Sample: Participants were first-year Arts (N = 616; 271 females, 345 males) and second-year Mathematics (N = 581; 241 females, 341 males) university students. Method: Students' epistemological beliefs were measured with the Schommer epistemological questionnaire (EQ, Schommer, 1990). Reflective thinking was measured with the reflective thinking questionnaire (RTQ, Kember et al., 2000). Student learning approaches were measured with the revised study process questionnaire (R-SPQ-2F, Biggs, Kember, & Leung, 2001). LISREL 8 was used to test two structural equation models--the cross-lag model and the causal-mediating model. Results: In the cross-lag model involving Arts students, structural equation modelling showed that epistemological beliefs influenced student learning approaches rather than the contrary. In the causal-mediating model involving Mathematics students, the results indicate that both epistemological beliefs and learning approaches predicted reflective thinking and academic performance. Furthermore, learning approaches mediated the effect of epistemological beliefs on reflective thinking and academic performance. Results of this study are significant as they integrated the three theoretical frameworks within the one study.