Two methodologies were developed to quantify and classify the content of cotton biodiesel in blends with diesel using medium infrared spectroscopy associated with partial least squares (PLS) and partial least squares discriminant analysis (PLS-DA). The PLS model, developed to determine the biodiesel content, was validated on the basis of some merit figures: selectivity, sensitivity, analytical sensitivity, limit of detection, limit of quantification and test for systematic error. The fit of this model was also evaluated using the correlation of current and predicted values of the calibration and prediction sets - a high correlation was observed, with correlation coefficient exceeding 0.99, and relatively low errors for the parameters. Qualitative monitoring was done using the PLS-DA model, whose efficiency was analyzed based on parameters of sensitivity and specificity. These parameters showed 100% correct classification in the samples used for calibration and prediction of biodiesel content in the Brazilian B10 fuel for diesel engines. The good results for application of the two models suggest that these analytical methodologies are feasible and efficient and can be used by inspection bodies for quantitative and qualitative control of this fuel. In addition, these methodologies are quick, of low cost and allow in-situ analysis with portable equipment.