We have modeled the airport surface as a network, with taxiways as links and their major intersections as nodes. The time required to travel over each link is random, and follows a set of stochastic processes. Our objective is to define a control strategy for this network, with the objective of minimizing aircraft surface fuel burn and emissions, as well as reducing takeoff delays. The approach that we have proposed has the advantage of explicitly accounting for uncertainty in surface operations, while still aiming to realize a large proportion of the potential benefits from congestion mitigation.