The use of virtual reality (VR) in formal education has burgeoned in recent years, with enthusiastic uptake by teachers and instructors across a wide range of subject areas and academic disciplines. We conducted a systematic meta-analysis of effects of VR on Science, Technology, Engineering, and Mathematics learning from middle school through postsecondary education. Eighteen published journal articles met inclusion criteria, yielding 52 effects from 2214 participants. VR has an overall positive effect on learning of g = 0.33, with the largest significant moderator effects for redesign of VR, classroom settings, science learning, desktop displays, and all types of learning outcomes (factual, conceptual, and transfer). Results depart somewhat from Howard's (2019) and Wu et al.'s (2020b) meta-analyses of VR across learning and treatment, multiple domains, and ages; in our study, desktop VR showed larger effects than head-mounted display, and we found positive effects for all learning outcome types. One trend within studies showing the largest effects is the inclusion of active learning techniques, which may shift learners' focus from interesting but irrelevant details to the most learning-relevant aspects of the VR learning environment.