The response of a Fourier-transform infrared (FTIR) instrument to changes in absorbance is inherently nonlinear for a number of reasons. One is that the interferogram acquired by the FTIR is truncated and then apodized before further processing of the data is accomplished. A commonly used apodization function in open-path FTIR research is triangular apodization, and all the research presented here has been done with that function. We calculated a set of absorption spectra by using the HITRAN database, covering ranges in both concentration and temperature for water, ammonia, and methane. Plots of these data reveal nonlinear results. The commonly used analysis technique, classical least squares, assumes that the response is linear. We describe some of the effects of this nonlinearity and present ways to address these effects.