This study explores whether and how the relationship between metacognitive monitoring and metacognitive control in self-regulated learning (SRL) is mediated by personal factors such as motivation, personal epistemology, metacognitive awareness, and other individual difference variables. An eye tracking system was used to accurately capture data pertaining to two aspects of metacognitive processes, monitoring and control, in SRL processes. These data were acquired while participants were engaged in studying and restudying some basic concepts of Number Theory with gStudy, a multi-featured learning software tool. This study yielded three significant findings. First, 37 of the 75 participants allocated more restudy time to information they judged not well learned, and 38 people allocated more restudy time to information they judged well learned. This result is not aligned with the dominant model of metacognition proposed by Dunlosky and Herzog (1998) which claims that learners allocate more study time to information they judge difficult. Second, three personal factors were statistically strongly associated with the relation between people’s judgments of learning and their allocations of restudy time: monitoring, achievement, and calibration. These represent two categories of learner characteristics: metacognitive awareness and achievement-related factors. Third, individual differences underlying self-reports about metacognitive control operations are fundamentally different from those underlying on-the-fly metacognitive control. This study reveals the dual-process character of online metacognition, different mechanisms of online metacognition from self-reported metacognition, as well as illuminates some limitations of self-report methodologies in measuring SRL in real time and the importance using state-of-the-art technologies in research on real-time SRL processes.