a b s t r a c tAn experimental and numerical investigation of the chemiluminescence signals from OH * , CH * , and CO 2 * is conducted for laminar premixed conical CH 4 /air flames diluted with CO 2 or N 2 . Experiments are conducted either at fixed equivalence ratio or fixed adiabatic flame temperature. An ICCD camera, equipped with different narrow bandpass filters, is used to record flame images at 307 nm (OH * ), 430 nm (CH * ), and 455 nm (CO 2 * ). A spectrometer is also used to correct the OH * and CH * emissions from the CO 2 * broadband background emission. Measured chemiluminescence intensities are then compared to one dimensional freely-propagating-flame direct simulations accounting for the chemistry of the excited radicals. Simulations predict accurately the OH * chemiluminescence intensity, independently of the diluent nature and concentration. Correction for the CO 2 * background has a weak influence on the recorded OH * signal. Predictions of CH * emissions are also in good agreement with experimental data if the CO 2 * background intensity is subtracted from intensity measurements. Measured and calculated CO 2 * emissions lead to acceptable results using a simplified chemistry mechanism for CO 2 * and an heuristic model for its emission intensity. Finally, it is shown that CO 2 dilution modifies chemiluminescence intensity couples and particularly the OH * /CO 2 * intensity ratio. These ratios regularly decrease with CO 2 dilution, a feature which is reproduced by the simulations. It is then shown that the ratio OH * /CO 2 * is well suited to infer the CO 2 diluent concentration in diluted CH 4 /air flames, a method which appears not feasible for sensing N 2 in N 2 -diluted CH 4 /air flames.