Increasingly, researchers have suggested the benefits of temporal analysis to improve our understanding of the learning process. Sequential pattern mining (SPM), as a pattern recognition technique, has the potential to reveal the temporal aspects of learning and can be a valuable tool in educational data science. However, its potential is not well understood and exploited. This chapter addresses this gap by reviewing work that utilizes sequential pattern mining in educational contexts. We identify that SPM is suitable for mining learning behaviors, analyzing and enriching educational theories, evaluating the efficacy of instructional interventions, generating features for prediction models, and building educational recommender systems. SPM can contribute to these purposes by discovering similarities and differences in learners' activities and revealing the temporal change in learning behaviors. As a sequential analysis method, SPM can reveal unique insights about learning processes and be powerful for self-regulated learning research. It is more flexible in capturing the relative arrangement of learning events than the other sequential analysis methods. Future research may improve its utility in educational data science by developing tools for counting pattern occurrences as well as identifying and removing unreliable patterns. Future work needs to establish a systematic guideline for data preprocessing, parameter setting, and interpreting sequential patterns.