This paper proposes a new Artificial Intelligence technique known as Harmony Search (HS) for optimal tuning of PID controllers for Load Frequency Control (LFC). The system proposed here is a three area with reheat thermal system containing nonlinearities represented by Generation Rate Constraint (GRC), Dead Band and Boiler Dynamics. The proposed system has been chosen especially to mimic typical system behaviour in actual operation. Four different generations (versions) of HS have been used for tuning the PID controllers. The closed loop response of the system using the optimized PID gains has been compared with another one tuned by Genetic Algorithm (GA). Simulation results significantly verify that the sample controller can favourable performance and contribute efficiently enhancing the system dynamic behaviour.