The boom in “N-reservoirs-and-multicascade” large-scale hydropower systems (N-M-LSHSs) poses tremendous challenges to the short-term peak-shaving (STPS) operation of the power grid, which includes modeling difficulties, complex constraint-handling and high time-consumption. A decomposition and iterative search strategy (DISS) for STPS operation of N-M-LSHS is proposed to overcome these challenges. First, a decomposition strategy based on adjustability is used to divide an N-M-LSHS into several independent hydropower subsystems (hydrosubs), and a model based on this strategy is formulated to restructure the coupling connections of the N-M-LSHS. Second, a method for generating an initial procedure is used to determine the power output of all hydrosubs, where mixed-integer linear programming (MILP) is selected to solve each hydrosub to enhance the search efficiency. Finally, an iterative search procedure is developed to converge to an acceptable solution within a suitable time frame by dynamically updating the peak-valley differences of the residual load (PVDRL). The DISS is tested for the eleven cascade hydropower stations on the Lancang River in Southwest China. The simulation results show that the DISS can effectively improve the solution efficiency by guaranteeing the accuracy requirements and the optimized PVDRL for the dry and wet seasons is reduced by 55.28 % and 75.51 %, respectively.