Abstract: Lubricating oils are usually produced by solvent extraction to separate aromatics in order to achieve the desired specifications and better quality products. Among the different properties of lubricating oils, density and refractive index are some of the most important properties which can both be used for petroleum fluid characterization. Predictions of density and refractive index for naphthenic oils during solvent extraction by DMSO obtained by the pseudo-component approach and the quadratic correlation were both examined. The pseudo-component approach is a method to predict density and refractive index from composition while the latter merely relates density to refractive index. Results indicated that the predictions yielded by the pseudo-component method were in good agreement with experimental data for naphthenic oils. And the use of a function of refractive index (FRI20) as a pseudo-component property remarkably improved n20 predictions for the naphthenic mixtures. However, the density and refractive index predictions obtained by the quadratic correlation exhibited significantly higher deviations for naphthenic oils than those for paraffinic oils. Thus a new modified correlation of the same functional form was proposed for naphthenic oils. The modification significantly improved predictions for naphthenic oils, which presented similar accuracy as the pseudo-component approach. And the previous correlation was still used for paraffinic oils. Additionally, effect of temperature on density and refractive index of naphthenic oils was examined. Results showed that the modified quadratic correlation was accurate for describing the relationship between density and refractive index of naphthenic oils at 20–90 °C. The temperature dependence of density and refractive index for the raffinates and the extracts could be accurately described by the thermal coefficients for saturates and aromatics, respectively. Regarding the refractive index variation of the extracts with temperature, the empirical equation was proved to be a better option compared with the method using the thermal coefficient for aromatics.