Head-mounted displays enable social interactions in immersive virtual environments. However, it is yet unclear whether the technology is also suitable for collaborative work between remote group members. Previous research comparing group performance in nonimmersive computer-mediated communication and face-to-face (FtF) interaction yielded inconsistent results. For this reason, we set out to compare multiuser immersive virtual reality (IVR), video conferencing (VC), and FtF interaction in a group decision task. Furthermore, we examined whether the conditions differed with respect to cognitive load and social presence. Using the hidden profile paradigm, we tested 174 participants in a fictional personnel selection case. Discussion quality in IVR did not differ from VC and FtF interaction. All conditions showed the typical bias for discussing information that was provided for all participants (i.e., shared information) compared with information that was only disclosed to individual participants (i.e., unshared information). Furthermore, we found that IVR groups showed the same probability of solving the task correctly. Social presence in IVR was reduced compared with FtF interaction; however, we found no differences in cognitive load. In sum, our results imply that IVR can effectuate efficient group behavior in a modern working environment that is characterized by a growing demand for remote collaboration.