A method to correct for fluctuating nonlinear background is suggested. The method is evaluated by experimental near-IR spectroscopy and by computer experiments that cover a wide range of possible variables. The current presentation of the method is suitable for monovariable analytical systems, such as mixtures in which the concentration of only one component is varied (and, as a result, the solvent's concentration changes accordingly). In spectroscopic applications, the concentration of an unknown sample can be determined from its spectrum and the spectra of two known reference samples, even though each individual spectrum has a different and nonlinear baseline. The method has been examined for polynomial baselines, up to the eighth degree. The method can handle extreme baseline fluctuations and is stable in a considerably wide range of the relevant parameters. Small deviations from monovariability cause only a slight (and linear) increase in analysis error, which means that the proposed algorithm is resistant to variable and unknown contamination.