Self-regulated learning is a crucial skill that may enable massive numbers of learners to thrive in MOOCs, but MOOC learners differ in their self-regulated learning skills, as low self-regulated learners need support to regulate their learning process in MOOCs. Designing self-regulated learning scaffoldings builds upon an accurate appraisal of learners' self-regulated learning and also a person-centered understanding of self-regulated learner profiles in MOOCs. Therefore, this study applied a two-parameter item response theory model to accurately evaluate online learners' self-regulated learning in MOOCs and then performed a cluster analysis to identify learner profiles in terms of their traits in each phase of self-regulated learning. The findings of this study identified five clusters of self-regulated learner profiles, upon which a further statistical analysis result indicated that the learners with lower forethought skills might be less committed to completing course assignments. Implications for self-regulated learning in MOOCs are discussed at the end.