Validly measuring aggression is challenging because self-reports are plagued with biased answer tendencies and behavioral measures with ethical concerns and low ecological validity. The current study, therefore, introduces a novel virtual reality (VR) aggression assessment tool, differentially assessing reactive and proactive aggression. Two VR tasks were developed, one in an alley environment (N = 24, all male, Mage = 23.88, 83.3% students) and an improved second one in a bar (N = 50, all male, Mage = 22.54, 90% students). In this bar VR task, participants were randomly assigned to either the reactive condition where they were triggered by a cheating and insulting dart-player or to the proactive condition where they could earn extra money by aggressing. Participants' level of self-reported aggression and psychopathy was assessed, after which they engaged in either the reactive or proactive VR task. Changes in affect and blood pressure were also measured. Aggression in the reactive VR task was evidenced to mostly display convergent validity because it positively correlated with self-reported aggression and total and fearless dominance factor scores of psychopathy, and there was a trend relationship with increased systolic blood pressure. The validity of the proactive aggression variant of our VR bar paradigm received less support, and needs more refinement. It can be concluded that VR is a potentially promising tool to experimentally induce and assess (reactive) aggression, which has the potential to provide aggression researchers and clinicians with a realistic and modifiable aggression assessment environment.