Statistical associating fluid theory (SAFT) is used to model CO 2 solubilities in single and mixed electrolyte solutions. The proposed SAFT model implements an improved mean spherical approximation in the primitive model to represent the electrostatic interactions between ions, using a parameter K to correct the excess energies ("KMSA" for short). With the KMSA formalism, the proposed model is able to describe accurately mean ionic activity coefficients and liquid densities of electrolyte solutions including Na + , K + , Ca 2+ , Mg 2+ , Cl -, Br -and SO 4 2-from 298.15 K to 473.15 K using mostly temperature independent parameters, with sole exception being the volume of anions.CO 2 is modeled as a non-associating molecule, and temperature-dependent CO 2 -H 2 O and CO 2 -Ion cross interactions are used to obtain CO 2 solubilities in H 2 O and in single ion electrolyte solutions. Without any additional fitting parameters, CO 2 solubilities in mixed electrolyte solutions and synthetic brines are predicted, in good agreement with experimental measurements.