The tracer diffusion coefficients, D-12, are fundamental properties for the design and simulation of rate-controlled processes. Nowadays, under the scope of the biorefinery concept and strict environmental legislation, the D-12 values are increasingly necessary for extractions, reactions, and chromatographic separations carried out at supercritical conditions, particularly using carbon dioxide. Hence, the main objective of this work is the development of accurate and simple models for the pure prediction of D-12 values in supercritical CO2. Two modified Stokes-Einstein equations (mSE(1) and mSE(2)) are proposed and validated using a large database comprehending extremely distinct molecules in terms of size, molecular weight, polarity and sphericity. The global deviations achieved by the mSE1 (Eqs. (2) and (13)) and mSE(2) (Eqs. (5), (13), (3), (4)) models are only 6.38% and 6.75%, respectively, in contrast to the significant errors provided by well known predictive correlations available in the literature: Wilke-Chang, 12.17%; Tyn-Calus, 17.01%; Scheibel, 19.04%; Lusis-Ratcliff, 27.32%; Reddy-Doraiswamy, 79.34%; Lai-Tan, 25.82%. Furthermore, the minimum and maximum deviations achieved by the new models are much smaller than those of the reference equations adopted for comparison. In conclusion, our mSE(1) and mSE(2) models can be recommended for the prediction of tracer diffusivities in supercritical CO2. (C) 2013 Elsevier B.V. All rights reserved.