In the hydrocarbons downstream business, it is very beneficial to quickly and reliably determine the physical and/or chemical properties of fuels. In this context, a nondestructive method was applied using midband Fourier transform infrared (FTIR) spectroscopy in association with multivariate partial least squares (PLS) chemometrics to determine the properties of nine groups of middle distillates (diesels) boiling in the range 180-370 degrees C. This method enables identification of one single diesel property at a time or a group of properties (32 properties) simultaneously in the spectral data'between 4000-650 cm(-1); with a minimum number of steps and without any sample preparation. The method was further used for two blends prepared from individual diesel samples. The results showed that using PLS models to process FTIR data is a practical analytical method to predict diesel fuel properties. Statistically, the results obtained showed low standard deviations, a very low root mean square error of cross-validation (RMSECV), low uncertainty values, less than 10 factors, but high correlation coefficient, R-2, and performance index (PI) values.