Electrolyte systems are of importance for a variety of applications from diverse fields, including thermal separation processes. Whereas various thermodynamic models for the correlation of properties of electrolyte systems have been developed, much fewer models allow predictions for such systems without the availability of experimental data. For this reason, a new model for electrolyte systems, which is based on the predictive thermodynamic model COSMO-RS, was developed in this work. The model was parameterized based on an extensive and diverse training set including liquid-liquid equilibrium data. Based on this parameterization, it could be shown that the model can also be applied to ions that were not part of the training set.