This work focuses on models suitable for taking into account the spectral properties of combustion gases in computationally demanding applications, such as computational fluid dynamics. One such model, which is often applied in combustion modelling, is the weighted-sum-of-grey-gases (WSGG) model. The standard formulation of this model uses parameters fitted to a wide range of temperatures, but only for specific ratios of H2O to CO2. Then, the model is limited to gases from fuels with a given composition of hydrogen and carbon, unless several sets of fitted parameters are used. Here, the WSGG model is modified to account for various ratios of H2O to CO2 concentrations. The range of molar ratios covers both oxy-fuel combustion of coal, with dry- or wet flue gas recycling, as well as combustion of natural gas. The non-grey formulation of the modified WSGG model is tested by comparing predictions of the radiative source term and wall fluxes in a gaseous domain between two infinite plates with predictions by a statistical narrow-band model. Two grey approximations are also included in the comparison, since such models are frequently used for calculation of gas radiation in comprehensive combustion computations. It is shown that the modified WSGG model significantly improves the estimation of the radiative source term compared to the grey models, while the accuracy of wall fluxes is similar to that of the grey models or better. (C) 2011 The Combustion Institute. Published by Elsevier Inc. All rights reserved.