High temperature solar-thermal reaction processes can be carried out within closed-cavity solar receivers in which concentrated solar energy enters the cavity through a small aperture or window and is absorbed either directly by reactants or by tubes containing reactant mixtures. Accurate modeling of radiation transfer phenomena in the solar receiver is critical for predicting receiver performance and improving receiver design. The accuracy of the finite volume (FV) method is evaluated in comparison to Monte Carlo (MC) techniques for both the concentrated solar energy and the energy emitted by heated surfaces in a receiver with either absorbing/diffusely emitting or specularly reflective cavity walls. Models are solved for two-dimensional slices of each of two receiver configurations with four spatial grids ranging from 2300 to 133,000 mesh elements, and three different angular grids. Solar radiative energy is described by a simplified uniform spatial profile at the receiver aperture that is either collimated or diffuse. Quantitatively accurate FV solutions for the solar energy either require highly refined angular and spatial grids, or are not possible on the mesh sizes investigated in this study. FV solutions for the emitted energy are sufficient even on coarse angular and spatial grids. FV solutions are least accurate when the cavity is highly specularly reflective or the absorber area is minimized, and tend to improve as the character of the incident solar energy changes from collimated to diffuse. Based on these results, a hybrid MC/FV strategy is proposed for use in combined radiation and convection/conduction heat transfer models. (C) 2012 Elsevier Ltd. All rights reserved.