Nationally, Departments of Motor Vehicles (DMV) license novice drivers based in part on on-road assessments. Intuitively it is assumed that such assessments are fair, reliable and valid measures of minimal driving competency. Upon further reflection, this would be difficult, given the subjectivity of a huge range of driving examiners that approach this assessment with different training backgrounds, life distractions and biases from examination to examination, the different road, traffic, lighting and weather conditions from one examination and DMV center to the next, and the minimal driving challenges in such assessments. For example, a typical on-road test involves only a 4 mile road segment with 2 left turns, 4 right turns, 1 lane change, pulling into a turn lane, and 1 speed limit change. It does not include highway driving nor defensive driving maneuvers. Additionally, such on-road assessments are both potentially dangerous and time demanding/expensive. A less expensive, safer, more challenging, objective, reliable, and valid procedure may be the use of Virtual Reality Driving Simulation (VRDS) that administers consistent and more extensive driving challenges to all examinees, which is evaluated in an objective manner based on normative data from current safe drivers. This presentation describes the experience and presents the data from a project where VRDSs were set up in two DMV facilities and a Research facility. The goals of this project were to determine whether VRDS assessments are just as reliable, discriminating and acceptable to the public as on-road assessments, and whether performance on the simulator predicts future on-road driving mishaps.