In this paper, a hybrid algorithm is developed by incorporating the egg-laying and immigration mechanisms of cuckoo optimization algorithm (COA) into harmony search (HS) algorithm (HSCOA), to design a secondary controller for two practical models of load frequency control (LFC) problem. Initially, a two-area non-reheat thermal power system is considered and the gains of PID and fuzzy PI/PID controllers are adjusted by the proposed tuning method. The superiority of HSCOA in regulating controller gains is demonstrated by evaluation and comparison of the obtained transient outcomes over some other published approaches in literature. To prove the satisfaction of the robustness in designed LFC by means of the proposed method, the performance of HSCOA based fuzzy PID controller is extensively verified under varying loading condition and some critical parameters related to the considered power plant. To add further practical challenge, the governor dead band (GDB) is included in the concerned system modeling to study the advantages of the HSCOA tuned fuzzy PID controller in handling the properties of nonlinearity in the system model. Time domain simulation of transient responses indicates that the designed controller operates satisfactorily to deal with the GDB nonlinearity and outperform other published techniques. Furthermore, to demonstrate the effective feasibility of the proposed method, the study is extended to a two-area multi-source power system with/without consideration of HVDC link. It is observed that HSCOA optimized fuzzy PID controller gives superior quality outcomes in comparison to other reported strategies. Finally, the robustness of the controller’s gains designed for the concerned power system is investigated under various scenarios of change in size, location and pattern of step load perturbation.