Real-time multisensory social interactions (RMSIs) between people are at the center of the metaverse, a new computer-mediated environment consisting of virtual “worlds” in which people act and communicate with each other in real-time via avatars. This research investigates whether RMSIs in the metaverse, when accessed through virtual-reality headsets, can generate more value for interactants in terms of interaction outcomes (interaction performance, evaluation, and emotional responses) than those on the two-dimensional (2D) internet (e.g., Zoom meetings). We combine theoretical logic with extensive field-experimental probes (which support the value-creation potential of the virtual-reality metaverse, but contradict its general superiority) to develop and refine a framework of how RMSIs in the metaverse versus on the 2D internet affect interaction outcomes through interactants’ intermediate conditions. The refined framework serves as foundation for a research roadmap on RMSIs in the metaverse, in which we highlight the critical roles of specific mediating and moderating forces along with interactional formats for future investigations of the metaverse and also name key business areas and societal challenges that deserve scholarly attention.