Operators of the Army's Unmanned Aircraft Systems (UAS) (Raven, Shadow, Hunter, and Gray Eagle) are all initially trained with a combination of simulated and live flight. The balance of live training on these systems currently ranges from 100% for Raven to 40% for Shadow. Requiring live flight is both expensive and can have a significant impact on training throughput. Given that UAS's are controlled using computer interfaces, it seems reasonable to ask whether live flight training is necessary at all. This paper examines the optimal balance between live and simulated initial operator training for Army UAS's. To do this, we examined the current programs of instruction, Training Aids, Devices, Simulators, and Simulations (TADSS), and we interviewed both students and instructors about training challenges they faced. It was clear from this effort that there is not a one size fits all answer for all UASs. Several factors were found to be critical determinants of an optimal balance including the capabilities of the TADSS, the instructors' ability to leverage the full capabilities of their TADSS, the cost of constructing or improving TADSS, the tasks required to be trained within the POI, and frequent changes to the UAS operational software. In developing these recommendations, we wanted to avoid reducing the quality of the training or shifting training from the institution to the unit. We concluded that the current TADSS for Shadow, Hunter, and Gray Eagle are good enough to reduce live flight training by about 10% to 20% depending on the UAS. This would save from 40 to 112 man-days of training time per class (assuming 20 student classes). Cost-benefit analysis of Raven training, on the other hand, indicated no benefit of reducing live flight training.