We present a scheme for run-to-run disturbance rejection in optimization-based feedforward path following of a remotely piloted aircraft system (RPAS). The proposed scheme is based on the inter-run estimation of unknown disturbances, such as wind forces and model uncertainties. These disturbance estimates are introduced in an optimal control problem used to compute feedforward controls. In order to achieve good run-to-run disturbance rejection, the structure of the underlying stabilizing flight control of the RPAS is taken into account. In this work, we consider flight control based on adaptive reference following and the special case of the unmanned helicopter ARTIS. We present simulation results and flight test data. These results underpin that the proposed approach significantly decreases flight path deviations in a run-to-run fashion.