The classical Malkmus statistical narrow-band model was extended with a fictitious gas method to improve the numerical accuracy of the infrared radiation signature of high-temperature gas in aeroengine exhaust systems. In this study, the accuracy of the extended model and the classical Malkmus statistical narrow-band model were evaluated. The results show that the numerical accuracy of the classical Malkmus statistical narrow-band model was improved significantly by the fictitious gas assumption, particularly for nonisothermal and nonhomogeneous gases. Compared with the line-by-line results, the root mean square error of the classical Malkmus statistical narrow-band model for the average band transmissivity of CO₂-H₂O-N₂ mixture is 0.018, while the root mean square error of the fictitious gas-based Malkmus statistical narrow-band model is 0.012, which is reduced by 33.3% compared with the former.