Multi-objective model predictive control (MOMPC) is a challenging task in two regards: firstly, the Pareto front approximation has to be provided under real-time constraints and, secondly, an automated decision-making has to be implemented. Thus, we propose a combination of the Pascoletti–Serafini scalarization method with high-order sensitivity computations to approximate the Pareto front at the current MPC step using polynomials. The decision-making is designed to guarantee stability of the MPC loop. In our numerical evaluation, this method outperforms a standard approach agnostic to sensitivity information both in quality of the result and computation time.