An in-core nuclear fuel management code for pressurized water reactor reload design has been developed that combines the stochastic optimization technique of simulated annealing with a computationally efficient core physics model based on second-order accurate generalized perturbation theory. The approach identifies the placements of feed fuel, exposed fuel with assembly orientations, and burnable poisons within the core lattice that optimize fuel cycle performance or thermal margin according to one of the following objectives: maximization of keff at a target end-of-cycle (EOC) burnup, minimization of the maximum radial power peaking over the cycle, or maximization of region average discharge burnup, and subject to constraints on radial power peaking, discharge burnup, and moderator temperature coefficient. Each objective examined for a typical cycle 2 reload indicated the existence of multiple optimal solutions. A comparison of the loading patterns obtained for the same fuel inventory shows that the marginal cost associated with achieving a 6.1% reduction in the maximum radial power peaking is equivalent to a 15.0% increase in fuel cycle costs for the specific core analyzed. Alternatively, an optimum loading pattern was found that increased the region average discharge burnup by 11.4% more than the one that maximizes the EOC keff, with the added expense of an increase in feed enrichment required to offset an otherwise 11.2% decrease in cycle length.