Background and context: There were few studies indicating if students' computational thinking (CT) self-efficacy and their CT performance were aligned with each other. Objectives: The study was to investigate if there was a discrepancy between students' CT self-efficacy and their CT performance. Method: Involving 104 non-CS undergraduate students as participants, this study employed descriptive statistics, one-sample tests, and prediction modules to perceive if a discrepancy existed. Findings: The findings revealed that students' CT self-efficacy and their performance on understanding CT showed an opposite pattern and failed to meet instructors' expectations. Moreover, students' CT self-efficacy did not significantly predict their CT performance on application or understanding. Implications: The detected discrepancy suggests that students' CT understanding needs to be improved. A variety of approaches should be employed to evaluate their CT self-efficacy. CT self-efficacy surveys need to be refined to enhance the validity of CT self-efficacy measurement.