Based on the maximum entropy principle used to reconstruct the neutron flux distribution in nodal calculations, an improved method that utilizes Lagrange multipliers and incorporates corner-point fluxes is described. The probability distribution that maximizes the entropy provides the most unbiased objective probability distribution within the partial information known. The flux distribution on the boundary of a fuel assembly is transformed into the probability distribution in the entropy expression. The most objective boundary flux distribution is then deduced by numerical evaluation of the Lagrange multipliers. This boundary flux distribution is used as the boundary condition in an embedded heterogeneous assembly calculation to provide the detailed flux distribution. The application of the new method to two pressurized water reactor benchmark problems shows that the reconstruction errors are very small. In particular, the minimum cross-entropy method, which is a generalization of the maximum entropy method, when tested on a benchmark problem with skewed fluxed distribution, provides much smaller errors than the existing form function method.