Purpose <p>The goal of this study was to determine whether residents can be taught to place spinal pedicle screws in a virtual reality (VR) environment and if the results are comparable to conventional training.</p> Methods <p>In this pilot study, four neurosurgery residents with less than one year of experience in spine surgery performed three rounds of pedicle screw instrumentation from L1 to S1. Both groups performed the first instrumentation after self-instruction. Before the second instrumentation, one group had conventional training, which was instruction on a synthetic bone model, and the other group was trained in a virtual environment. For the last instrumentation, both groups had both trainings. After completion, screw placement was rated by a blinded reader according to the trajectory and integrity of the screw canal. For statistical analysis, we used logistic regression for individual variability with Firth’s bias reduction method.</p> Results <p>After self-instruction 21/48 (44%) screws were misplaced, 6 (25%) in the group before conventional training and 15 (64%) in the group before VR-training. Both trainings led to a significant reduction of screw misplacement, with only a single misplaced screw in the VR-training group (2%, <i>p</i> &lt; 0.001) and none in the conventional training group (<i>p</i> = 0.008). In both groups, no significant difference was found regarding the misplacement rate from the first to the second training (<i>p</i> = 0.19).</p> Conclusions <p>Training in a virtual reality environment for lumbosacral pedicle screw placement is associated with an easy learning curve. Comparable outcomes are observed with conventional training, however, definitive conclusions regarding the relative effectiveness of either approach will require validation in studies with larger cohorts.</p>