A selective algorithm for nondispersive infrared (NDIR) sensing of gases with overlapping absorption spectra was developed and evaluated in modified multichannel NDIR sensor. Measurements in the optic channel with the spectral band where two gas species (target and secondary gas) have overlapping absorption lines are complemented by additional measurements in second channel where spectral absorption for only one gas (secondary gas) is present. The real concentration for the target gas is retrieved by adjusting the absorption data obtained in the overlapping gas spectra's optic channel, with respect to the absorption data retrieved in the second optic channel that has sensitivity only for the secondary gas. An implementation example is performed by obtaining the true concentration of CH4 (as target gas) in a mixture with H2O vapor. The channel for the target gas is equipped by an optic filter with spectra at 3.375 μm where both CH4 and H2O have absorption lines. The complementary second channel provides sensing in spectra at 2.7 μm where only H2O have absorption. Data from a third channel, at 3.95 μm, is used as reference value for 'zero-level' calibration. A calibration procedure was developed and tested, which involves matching of the absorbed light energy in target and secondary channels in humid reference environments. A selective algorithm for sensing of CH4 with elimination of spectral impact from H2O was validated in environments with variable CH4 and H2O concentrations. By implementing the multispectral approach and the developed algorithm, an uncertainties of 5-10 ppm relative the reference concentrations were achieved. For the environments where selective algorithm was validated this should be compared to an uncertainty of 70-90 ppm for the non-corrected CH4 concentration.