Cavity receivers used in solar power towers and dish concentrators may lose considerable energy by natural convection, which reduces the overall system efficiency. A validated numerical receiver model is desired to better understand convection processes and aid in heat loss minimization efforts. The purpose of this investigation was to evaluate heat loss predictions using the commercial computational fluid dynamics (CFD) software packages fluent 13.0 and solidworks flow simulation 2011 against experimentally measured heat losses for a heated cubical cavity receiver model (Kraabel, 1983, “An Experimental Investigation of the Natural Convection From a Side-Facing Cubical Cavity,” Proceedings of the ASME JSME Thermal Engineering Joint Conference, Vol. 1, pp. 299–306) and a cylindrical dish receiver model (Taumoefolau et al., 2004, “Experimental Investigation of Natural Convection Heat Loss From a Model Solar Concentrator Cavity Receiver,” ASME J. Sol. Energy Eng., 126(2), pp. 801–807). Simulated convective heat loss was underpredicted by 45% for the cubical cavity when experimental wall temperatures were implemented as isothermal boundary conditions and 32% when the experimental power was applied as a uniform heat flux from the cavity walls. Agreement between software packages was generally within 10%. Convective heat loss from the cylindrical dish receiver model was accurately predicted within experimental uncertainties by both simulation codes using both isothermal and constant heat flux wall boundary conditions except when the cavity was inclined at angles below 15 deg and above 75 deg, where losses were under- and overpredicted by fluent and solidworks, respectively. Comparison with empirical correlations for convective heat loss from heated cavities showed that correlations by Kraabel (1983, “An Experimental Investigation of the Natural Convection From a Side-Facing Cubical Cavity,” Proceedings of the ASME JSME Thermal Engineering Joint Conference, Vol. 1, pp. 299–306) and for individual heated flat plates oriented to the cavity geometry (Pitts and Sissom, 1998, Schaum's Outline of Heat Transfer, 2nd ed., McGraw Hill, New York, p. 227) predicted heat losses from the cubical cavity to within experimental uncertainties. Correlations by Clausing (1987, “Natural Convection From Isothermal Cubical Cavities With a Variety of Side-Facing Apertures,” ASME J. Heat Transfer, 109(2), pp. 407–412) and Paitoonsurikarn et al. (2011, “Numerical Investigation of Natural Convection Loss From Cavity Receivers in Solar Dish Applications,” ASME J. Sol. Energy Eng. 133(2), p. 021004) were able to do the same for the cylindrical dish receiver. No single correlation was valid for both experimental receivers. The effect of different turbulence and air-property models within fluent were also investigated and compared in this study. However, no model parameter was found to produce a change large enough to account for the deficient convective heat loss simulated for the cubical cavity receiver case.