The conventional calculation of power peaking near water gaps assumes an abrupt change in the neutron spectrum at a gap-core interface. The assumption can be seriously in error, and can result in discrepancies of 50% between calculated and experimental peaking values. In this paper, a position-dependent spectrum is obtained by the use of diffusion theory which, when used in peaking calculations, reduces the discrepancy between theory and experiment to the order of 5–10% or less. Recipes based on the position-dependent spectrum are obtained for the specification of position-dependent cross sections which may be used in standard diffusion theory codes. The use of these cross sections in the codes results in an estimate of power peaking factors which represents a considerable improvement over the results given by conventional calculations.