In STEM education, a thorough understanding of the interaction of self-efficacy and metacognitive monitoring behaviors is needed to refine theories and inform the design of instructional supports for students with varying levels of motivation and self-regulation skills. We examined how students' (n = 1063) exam scores in an undergraduate life science course were influenced by their self-efficacy and online metacognitive monitoring behaviors by integrating variable-centered and person-centered approaches. In a semester-long study, students' self-efficacy judgements made at the end of the semester were stronger predictors of students' final exam performance than those made at the beginning of the semester. Results further suggested that the influence of self-efficacy on exam scores decreased as online monitoring behaviors increased. Students’ prior GPA predicted membership in three latent profiles indicated by 1) high self-efficacy with high metacognitive monitoring activity; 2) high self-efficacy with low metacognitive monitoring activity; and 3) low self-efficacy with low metacognitive monitoring activity. Learners with high self-efficacy and high monitoring activity outperformed those with high self-efficacy and low monitoring, who outperformed those with low self-efficacy and low monitoring on exams.