A stagewise optimization of the refueling schedule for light-water reactors has been developed with emphasis on the nuclear model. The decision variables to be determined are end-of-cycle (EOC) reactivity distribution, energy output, power distribution, number of fresh fuel assemblies, number of reinsertion of used assemblies, selection of assemblies for discharge, and allocation of each fuel assembly in a two-dimensional core geometry. Division of the total problem into six phases permits usage of the most effective method in each phase. This study employed such techniques as linear programming for regionwise shuffling optimization, linear iterative search for the optimal EOC state, the minimum integrated k-deviation method for a guess allocation, and direct search for the optimal allocation of each fuel assembly, etc., all of which are interrelated. The applicability of this method to a commercial light-water reactor was demonstrated for a 1300-MW(th) boiling-water reactor by successfully generating a ten-cycle refueling schedule using a fixed enrichment of initial and reload fuel and allowing reinsertion of discharged fuel assemblies from the first to the third cycles. The results indicate a savings of as much as 14% of the fresh fuel consumption over a conventional mixed four- and five-batch scatter loading, with thermal characteristics well within design limits.