Abstract This paper introduces an optimisation based control framework for autonomous helicopters. The framework contains a high-level model predictive control (MPC) and a low-level linear controller. The proposed MPC works in a piecewise constant fashion to reduce the computation burden and to increase the time available for performing online optimisation. The linear feedback controller responds to fast dynamics of the helicopter and compensates the low bandwidth of the high-level controller. This configuration allows the computationally intensive algorithm applied on systems with fast dynamics. The stability issues of the high-level MPC and the overall control scheme are discussed. Simulations and flight tests on a small-scale helicopter are carried out to verify the proposed control scheme.