The main purpose of this study was to examine the effects of Blended learning (BL) on undergraduate students' Conceptual Understanding (CU), engagement (SE), and epistemological beliefs (EBs) in learning Chemical Bonding and Molecular Geometry (CBMG) concepts. To achieve this goal, a mixed research approach was employed. For the quantitative approach, a pretest-posttest non-equivalent control group quasi-experimental design was employed. For the qualitative phase, a phenomenological design was used. The study was conducted at Bahir Dar University in the Amhara National Regional State of Ethiopia. A total of 86 undergraduate students (Male = 59 and Female = 27) participated, with the intervention group (IG; n = 42) taught via the BL approach and the comparison group (CG; n = 44) taught via the Conventional Teaching Method (CTM). A total of nine participants, six from IG and 3 students from CG, participated in the semi-structured interviews. The Three-tier Conceptual Understanding Diagnostic Test (TTCUDT), Student Engagement Scale (SES), and Colorado Learning Assessment and Science Survey (Class-Chem) were administered to both groups as pretest and posttest. Moreover, semi-structured interview questions were administered regarding CU, SE, and EBs after the treatment. The study employed quantitative and qualitative data. Quantitative data were analyzed using IBM SPSS Statistics version 26 and were examined using descriptive and inferential statistics, while qualitative data were analyzed thematically. Findings from both approaches indicated that students who engaged with the BL approach demonstrated greater CU, SE, and more sophisticated EBs than their peers who experienced CTM while learning CBMG concepts. Furthermore, the findings revealed that the BL approach significantly reduced students' MCs and enhanced their confidence. However, no statistically significant gender differences were observed in CU, SE, and EBs within either the intervention or comparison groups. The BL approach has been shown to substantially enhance students‘ CU, SE, and EBs regarding scientific knowledge acquisition. Furthermore, the findings of this study revealed that, within the BL environment, students‘ CU was significantly correlated with their engagement and EBs; moreover, increased SE contributed to the development of more sophisticated EBs. Although the implementation of BL has a significant positive effect on students' CU, SE, and EBs related to knowledge, various challenges were encountered in the learning environment due to technological, pedagogical, student-related, and institutional factors. Adopting carefully designed iii BL approaches enables educational institutions to improve teaching effectiveness and SE, while contributing to the growing evidence on the pedagogical impact of BL in Science, Technology, Engineering, and Mathematics (STEM) education. Drawing on these findings, further implications for practice and recommendations for future research are advanced