The partial least-squares (PLS) method was applied to predict the content of biodiesel in diesel using distillation curves (ASTM D86). Furthermore, a comparison between the regression coefficients of all algorithms was carried out, subsequently to the selection of the number of latent variables according to each prediction. The low values obtained for RMSEP and high correlation coefficients between real and predicted values indicated that the PLS was efficient in determining the biodiesel content. The model built contains diesel samples of several different compositions, replicating the diversity of fuels of the Brazilian market. In addition, the proposed method is inexpensive and is easy to implement, since it uses the results of a routine assay performed to evaluate the quality of the diesel used in road transport.