Web-based inquiry learning provides opportunities for students to take responsibility to regulate their learning. However, due to a lack of science inquiry-specific self-regulated learning (SRL) frameworks, there is insufficient understanding of SRL processes in inquiry-based science learning. This study aims to explore students' SRL patterns by using a comprehensive framework that combines SRL with the science inquiry process. Additionally, log-file data collected in the online science inquiry learning session were used to analyze students' SRL patterns. The results of the latent class analysis revealed four types of SRL learners: disengaged learners, proficient SRL learners, aimless reflective learners, and less reflective learners. Furthermore, we found significant differences in science achievement tests among different SRL learners. Specifically, proficient SRL learners and less reflective learners scored significantly higher than the other two types of learners. A difference was also found between proficient SRL learners and disengaged learners in terms of their self-determined motivation. Understanding the heterogeneity of SRL processes among students revealed from distinct SRL patterns informs how to provide targeted intervention and support for students who encounter difficulties in inquiry-based science learning.