A linear solvation energy relationship (LSER) method was used to develop a predictive model for the diffusivities of organic solutes in supercritical CO2 at infinite dilution. The LSER model was based on the diffusivities of 18 solutes and 104 data points for SC-CO2 in the range of 32-60 degrees C and 8-100 MPa. The independent variables in the model were empirically determined descriptors of the solute molecules and the dipolarity/polarizability Of CO2 at a given density. The model was tested for prediction accuracy by using the diffusivities of 10 solutes not included in the database. The model provided relative deviations less than 10% in the correlation and prediction of the diffusivities in supercritical CO2 of the organic solutes considered. The accuracy of the proposed LSER model is comparable with He-Yu [31] equation, which is the most effective correlation in the literatures. The coefficients of the model show that diffusivity in supercritical CO, is strongly dependent on the dipolarity/polarizability of CO2. The hydrogen-bond basicity and dipolarity/polarizability of the solute have significant effects on the diffusivity, whereas the hydrogen-bond acidity and excess molar refractivity are less important. The logarithm of the solute's gas-to-hexadecane partition coefficient used to model dispersion interactions and cavity formation processes was found to be statistically insignificant. (c) 2004 Elsevier B.V. All rights reserved.