Providing quality undergraduate STEM instruction in the twenty-first century is both a national priority and a continued pedagogical challenge. Over half a century of research has endorsed the use of cooperative learning–a form of active learning whereby small groups of students work interdependently in order to maximize all students’ learning–over didactic or competitive instructional designs that are typical of undergraduate STEM teaching. In this chapter, I review the evidence for cooperative learning in undergraduate STEM learning contexts alongside a discussion of key questions in cooperative learning research. Chiefly, it remains unclear whether students must be grouped homogeneously (all similar ability levels) or heterogeneously (mixed abilities) to produce achievement gains. Towards this question, I review key methodological considerations of the extant literature (conflation of performance and learning measures) and relatively underconsidered outcomes of cooperative learning (students’ metacognitive and social-affective changes) that are related to achievement in cooperative settings. Finally, I summarize results from a recent experiment conducted by my team that addresses these questions in an undergraduate introductory statistics context. I conclude the chapter with suggestions for classroom implementation and a call for future directions.