The evolution of a systematic procedure for fuel cycle optimization and a new approach using a linear program for optimal fuel allocation is presented. The process of loading, operating, and refueling a power reactor is viewed as a multistage decision process where each stage or cycle corresponds to a partial refueling of the core and the subsequent operating period, and the objective to be minimized is unit fuel cost. In the overall optimization procedure, k∞ is the state variable of the process and the power control problem is separated from the optimization process. Hence, precalculated results can be used in relating the attainable exposure distribution in a cycle and the resulting power distribution to the k∞ distribution at end of cycle. Interactive graphics have considerable merit in searching for feasible k∞ distributions. Dynamic programming has been applied to this process for certain limited loading schemes. The complexity of the decision vector with a general fuel location matrix led to abandonment of this approach and development of a linear program for optimal fuel allocation. The linear program selects reloading patterns for a few-region core model which minimize the present worth weighted total fuel cost, subject to the constraints of required region fuel loading and k∞ for each stage. It is sufficiently fast to be used as a subroutine in a systematic search for the global optimum of the remaining parameters of the decision vector, such as cycle period, region exposure, region k∞, and fuel enrichment. An iterative overall fuel cycle optimization procedure is outlined where coefficients of the reactor model in the linear program are modified by the results from higher level burnup programs.