ABSTRACT Research purposes and design This study aimed to compare the effects of different explicit approaches used in modeling-based instruction on students’ understanding of scientific models and modeling (USMM). In Study 1, a self-study approach was compared to a preliminary explicit approach. In Study 2, preliminary explicit approach and embedded explicit approach were compared. Methods For the data analysis, t tests and analysis of covariance (ANCOVA) were used to compare students’ USMM under the conditions of using different explicit approaches. Findings Results showed that in the first study, no statistically significant difference was found between self-study approach and the preliminary explicit approach in any of the USMM subscales. In the second study, students who learned with the embedded explicit approach had better understandings in all subscales of USMM, but not in the changing nature of scientific models (CNM) subscale, than those who learned with the preliminary explicit approach. Through qualitative analysis of the data, categories and examples of the teacher’s embedded epistemic guidance were illustrated. Suggestions for future professional development and research instrument design are provided.