Past studies have witnessed the wide application of assisted history matching for the calibration of dynamic reservoir models. Although the proposed algorithms have the potential to improve the history matching process in some synthetic cases, most of them have failed or have partially succeeded when applied to real, complex reservoirs. Thus far, identifying the most efficient optimization strategy for history matching has remained a challenging topic for research. In this paper a sequential approach is adopted whereby a reservoir model is replaced by a proxy model and multi-objective optimization algorithms are applied on misfit functions which was defined by combination of the proxy models and historical data. The proposed approach was tested on a case study involving a benchmark synthetic reservoir model with 14 years of production data. The data was freely provided by Imperial College London. The effectiveness of using individual optimization algorithms were quantified by using normalized root mean square error. The proposed approach is found to be efficient, robust and flexible. © IEOM Society International. © IEOM Society International.