These 2 studies investigate the extent to which an Embodied Mixed Reality Learning Environment (EMRELE) can enhance science learning compared to regular classroom instruction. Mixed reality means that physical tangible and digital components were present. The content for the EMRELE required that students map abstract concepts and relations onto their gestures and movements so that the concepts would become grounded in embodied action. The studies compare an immersive, highly interactive learning platform that uses a motion-capture system to track students' gestures and locomotion as they kinesthetically learn with a quality classroom experience (teacher and content were held constant). Two science studies are presented: chemistry titration and disease transmission. In the counterbalanced design 1 group received the EMRELE intervention, while the other group received regular instruction; after 3 days and a midtest, the interventions switched. Each study lasted for 6 days total, with 3 test points: pretest, midtest, and posttest. Analyses revealed that placement in the embodied EMRELE condition consistently led to greater learning gains (effect sizes ranged from 0.53 to 1.93), compared to regular instruction (effect sizes ranged from 0.09 to 0.37). Order of intervention did not affect the final outcomes at posttest. These results are discussed in relation to a new taxonomy of embodiment in educational settings. We hypothesize that the positive results are due to the embodiment designed into the lessons and the high degree of collaboration engendered by the co-located EMRELE.