The aim was to identify what epistemological beliefs: Source, utility, stability or nature of knowledge, typify students with Low and High Academic Performance when their beliefs are reflexive or naive. Taha Mohamed and El-Habbal (2013) provided conflicting data: students with high academic performance showed more naive epistemological beliefs than their peers with lower performance .381 participants from across the country were selected. Four extreme samples were created: a) students of reflexive epistemology and high academic performance (N = 95), b) reflexive epistemology with low academic performance (N = 94), c ) naive epistemology with high academic performance (N = 96) and d) naive epistemology with low academic performance (N = 96).Academic achievement tests were applied (before and after), and an introductory course in perception and learning,  and self-report inventories of Cognitive and self-regulatory Study Strategies and of Personal Epistemology.  Theoretical models were tested by SEM. The results showed that empirical models fitted with theoretical models, both statistically and practically: a) for reflexives of high  performance  was 176.43, p = .37, comparative fit index (CFI) = 0.97 and RMSEA = 0.02); b) for reflexives of low performance: c2 = 195.03 p = .39 CFI = 0.98 RMSEA = 0.02, while c) naive of high performance: chi2 = 95.49 p = .91 CFI = 1.0 RMSEA = 0.00 d) naive of low performances (chi2 = 120.67 p = .94 CFI = 0.00 RMSEA = 1.0). An interesting result shows that even if the reflexive epistemological beliefs are necessary components, they are not sufficient to support a successful academic performance. In models with naive beliefs is possible to find cases with high academic achievement and reflexive models may have low performance. The importance of cognitive and self-regulatory strategies as a differential factor in academic performance is confirmed and the contradictory effect is explained.