The method of loci (MoL) is a well-known mnemonic technique in which visuospatial spatial environments are used to scaffold the memorization of non-spatial information. We developed a novel virtual reality-based implementation of the MoL in which participants used three unique virtual environments to serve as their "memory palaces." In each world, participants were presented with a sequence of 15 3D objects that appeared in front of their avatar for 20 s each. The experimental group (N = 30) was given the ability to click on each object to lock it in place, whereas the control group (N = 30) was not afforded this functionality. We found that despite matched engagement, exposure duration, and instructions emphasizing the efficacy of the mnemonic across groups, participants in the experimental group recalled 28% more objects. We also observed a strong relationship between spatial memory for objects and landmarks in the environment and verbal recall strength. These results provide evidence for spatially mediated processes underlying the effectiveness of the MoL and contribute to theoretical models of memory that emphasize spatial encoding as the primary currency of mnemonic function.