This paper presents a feasible methodology to quantify and classify the mafurra methyl biodiesel content in a diesel blend using Fourier transform mid-infrared spectroscopy (FT-MIR) in combination with partial least squares (PLS) and partial least squares discriminant analysis (PLS-DA) and variable selection (siPLS and siPLS-DA). When the spectra were divided into 30 equally spaced intervals, the models combining four intervals (that is, si4PLS30 and si4PLS-DA30) showed statistically better results than PLS and overall PLS-DA. This comparison was performed on the basis of the application of F test values. The si4PLS30 model was validated on the basis of several figures of merit, and strong correlations between the actual concentration and forecasted values of the calibration and prediction sets, with a correlation coefficient higher than 0.99, were observed. The efficiency of the si4PLS-DA30 model was evaluated on the basis of sensitivity and specificity, and these parameters showed values of 1, indicatin...