Receding horizon optimal control is a special case of model predictive control in which optimal controls are computed based on predicted process conditions over some suitable time horizon. The controls are regularly recomputed in light of new measurements from the plant or changes in predicted conditions. It is important to allow for state constraints in this computation and we present a new algorithm which deals with this problem for systems described by DAEs of any index. It is also important to take account of uncertainty in both the model and the predicted inputs and we discuss possible approaches to deal with this.