This paper presents a decentralized model predictive based load frequency control for multi-area power systems. An MPC controller is designed for each area to decrease the effect of load disturbances and parameter uncertainties owing to variations in power system’s parameters. Also to estimate load changes, Fast Output Sampling (FOS) method is used. The estimated signal is defined as a measured disturbance for MPC controller. The MPC method uses feed-forward control strategy for rejecting load disturbance effect in each power system area. A three-area power system has been considered as a test case to evaluate the proposed control scheme’s performance. The results have been compared with a recent proposed robust LMI based PI control strategy. This comparison confirms that the proposed method has better performance than the LMI based PI control design in the presence of disturbances and uncertainities so that the frequency deviation and power flow changes between areas are effectively damped to zero with small oscillations in a short time.