This paper presents a finite control set model predictive strategy to control the three-phase four-leg neutral-point-clamped (NPC) inverters. The four-leg NPC inverter delivers power to the unbalanced/nonlinear three-phase loads, and it can produce three output currents independently. The proposed method uses the discrete-time model of the inverter and load to predict the load current and capacitor voltages behavior for each valid switching state of the inverter. The control method chooses a state which generates the minimum error between the output currents and their references and also between the capacitor voltages. The neutral-leg switching frequency reduction algorithm is also proposed to improve the efficiency of the converter. Through computer simulations, the feasibility of the proposed predictive control scheme is verified; it performs well, showing a clear capacity to compensate disturbances while balancing the DC-link capacitor voltages and reducing the average switching frequency.