The present paper builds on the literature that emphasizes the importance of self-regulation for academic learning or self-regulated learning (SRL). SRL research has traditionally focused on count measures of SRL processing events, however, another important measure of SRL is being recognized: time-on-task. The current study captures the influence of time spent on learning performance. We study time-on-task--from the perspective of self-regulated learning--in the context of clinical reasoning in an intelligent tutoring system. Specifically, we examine the link between the time spent in the three phases of SRL (forethought, performance, and self-reflection) and confidence in diagnosis and diagnosis correctness. Our analyses revealed non-significant links between the time spent in the three phases of SRL and diagnosis correctness. On the other hand, significant associations were found between the time spent in the three phases of SRL and confidence in diagnosis (confidence in diagnosis was: positively associated with time spent in forethought phase; negatively associated with time spent in performance phase; and, positively associated with time spent in self-reflection phase). In addition, confidence in diagnosis was positively linked to diagnosis correctness. Considering learning time offers an alternative perspective on regulation of learning and problem-solving performance. We conclude by offering implications of our findings and recommendations for further research.