A one-dimensional method based on a combination of the “nodal equivalence theory” and response matrix homogenization methods was previously described for determining environment-insensitive equivalent few-group diffusion theory parameters for homogenized radial reflector nodes of a pressurized water reactor. This reflector model, called the NGET-RM model, yields equivalent nodal parameters that do not account for the two-dimensional structure of the baffle at core corners; this can lead to significant errors in computed two-dimensional core power distributions. A semi-empirical correction procedure is proposed for reducing the two-dimensional effects associated with this particular one-dimensional reflector model. Numerical two-group experiments are performed for a given reflector configuration (and soluble boron concentration) to determine optimal values for the two empirical factors defined by this model. It is shown that the resultant factors are rather insensitive to core configuration or core conditions and that their application yields improved two-group NGET-RM reflector parameters with which accurate nodal power distributions can be obtained. The results are also compared with those obtained with another one-dimensional environment-insensitive model that has an extra degree of freedom utilized here to reduce two-dimensional effects. Some practical aspects related to the application of the proposed correction procedure are briefly discussed.