Load frequency control (LFC) in power systems is one of the most important issues in the field of optimizing power system performance that attracted the attention of many researchers. In this work, a sliding mode based load frequency control is developed on a three-area interconnected power system. The power system contains non-reheat, reheat, and hydraulic turbines which are distributed in these three areas respectively. Both governor dead band and generation rate constraint are included in the model of this power system. Our control goal is to regulate the frequency error, tie-line power error and area control error despite the presences of external load disturbance and system uncertainties. Additionally, an optimization process is proposed for optimal adjustment of the sliding mode control parameters of each of the three areas, aimed at improving integral performance and step response characteristics such as amount of overshoot, steady state error, and sitting time. The optimization was performed using the particle swarm optimization (PSO) algorithm, which is one of the most powerful algorithms for nonlinear problems solving. The sliding mode based load frequency controller is simulated on this three-area interconnected nonlinear power system. The simulation results verify the effectiveness of the sliding mode controller. In addition, the performance of SMC is compared with an optimized PI controller. The comparison study shows the superiority of the SMC to the PI controller in term of control performance. They also demonstrate the robustness of the sliding mode controller against parameter variations and external disturbances.