A prediction model based on the processing of FTIR spectra and partial least squares regression (PLS) was developed for the determination of thehydroperoxide number of diesel fuels. The sets of calibration and validation standards were composed of fresh and aged diesel fuels. The hydroperoxide number determined via the standard titration method ranged from 0 to 65 mg·kg-1. While the calibration standards were utilized for the model construction, the validation standards were used for its optimization and validation. Several preprocessing methods, together with various numbers of latent variables, were utilized to improve model prediction ability. The model with the lowest Root Mean Square Error of Prediction was developed using mean centering, variance scaling, second derivative, and smoothing methods. Both examined smoothing techniques, i.e., Savitzky-Golay and Gap-Segment derivative, provided similar results. The use of the commonly available and affordable FTIR method, allowing rapid analysis, proved to be cost effective alternative to highly erroneous and laborious titration methods utilizing toxic reagents. In general, the developed model showed good predictive ability and is a perfect solution for fast screening of oxidative aging level of conventional hydrocarbon-based fuels.