With increasing autonomy in unmanned vehicles, there is potential to field large numbers of them in distributed teams with a smaller number of human operators. The effects of human interaction with automation are often unintuitive and may complicate the design of such systems. In order to provide systems engineers with reliable early predictions of system-of-systems performance, a model and simulation of a multi-UAV, multi-operator surveillance system is developed in this paper. UAVs are modeled as agents, and humans are modeled as event queues. The model of human performance dynamically adjusts for fatigue, workload, and task difficulty. The combined models establish a platform for system capability prediction as a function of human team size, UAV team size, mix of diverse UAV types, hardware quality, and level of autonomy.