Nondispersive infrared (NDIR) CO₂ analyzer could produce erroneous CO₂ mole fraction measurements for an air sample when CO₂-in-air mixtures having different isotopic compositions than atmospheric CO₂ are used as the NDIR calibration gases. This is because (1) an optical band-pass filter equipped in a typical NDIR analyzer to minimize the interference effect from the other infrared-active species is basically designed to transmit only the absorption band of ¹²C¹⁶O₂ and (2) absorption bands for the other CO₂-related isotopologues, for example, ¹³C¹⁶O₂, are shifted to lower wave numbers depending on their isotope effects. To evaluate the effect of the isotopic composition on the NDIR response, we computed the theoretical relative molar response of the instrument to each isotopologue based on the infrared absorptance by the individual isotopologues. We then prepared a gravimetric ¹³CO₂-in-air mixture with CO₂ mole fraction of 380 ppm to experimentally determine the optical filter property. The apparent mole fractions of the ¹³CO₂-in-air mixture determined by three NDIR analyzers used in this study were 46, 94, and 27 ppm, indicating that the optical filters in these instruments substantially reduced the response to ¹³C¹⁶O₂. Based on these theoretical and experimental analyses, we evaluated the apparent difference in the CO₂ mole fraction determined by the three NDIR analyzers from the true value when isotopically lighter CO₂-in-air mixtures (δ ¹³C = −32.4‰ and δ ¹⁸O = +11.7‰), as compared to atmospheric CO₂ (δ ¹³C = ∼−8‰ and δ ¹⁸O = ∼+40‰), are used as calibration gases. The estimated difference varied with NDIR analyzers, ranging from −0.04 to −0.08 ppm for air samples with 380 ppm CO₂.