The modular multilevel converter (MMC) has found various applications in the high voltage industry during the recent years. the MMC consists of several identical submodules (SM). For proper operation of the MMC, balance must be maintained between the upper and lower arm sum capacitor voltages and the ac components of the circulating current must be suppressed. Since the equations which describe the dynamics of the MMC are highly interrelated, the MMC control becomes complicated. In this paper, a control method is proposed to achieve stable and balanced MMC control while reducing the ac components of the circulating current as much as possible. The proposed technique is based on weighted model predictive controlwhich uses genetic algorithm to produce the optimum upper and lower arm insertion indices. The algorithm then produces the corresponding switching patterns by the use of submodule sorting algorithm. The weighting factors were selected based on simulation results. Simulations were carried out on a single-phase MMC with 10 submodules with possible extension to higher levels and phases. Theoretical analysis and simulation results are presented and discussed. They all confirm the effectiveness of the proposed method.