This study used both questionnaire and clickstream data to examine the relations between study approach, study time, and academic performance in low-stake and high-stake assessments in flipped classrooms. The questionnaire data identified two clusters of students reporting deep and surface study approaches. Students who reported a deep study approach performed better than those who reported a surface study approach on both low-stake and high-stake assessments. The clickstream data detected two clusters of students using either preparation-oriented or assessment-oriented approaches. Students who adopted preparation-oriented approach achieved higher on both low-stake and high-stake assessments. Students’ study time measured by both questionnaire data and clickstream data consistently contributed to both low-stake and high-stake assessments. However, the contributions of study approach to academic performance were not consistent across the two types of data, as self-reported study approach was only a significant predictor of high-stake but not low-stake assessment, suggesting that study approach measured by questionnaire data might not be sufficiently sensitive to detect variations in study approaches in flipped classrooms. Pedagogically, the results suggest that raising students’ awareness of the importance of pre-lecture learning and aligning assessments and learning activities may foster quality flipped classroom learning. Methodologically, the study demonstrates the advantages of combining questionnaire and clickstream data to understand students’ experiences of learning in flipped classrooms.