Several methods for the description of the retention behavior in micellar liquid chromatography (MLC) were described previously. Thereby, the most common are the linear solvation energy relationships (LSER). However, for the evaluation of the LSER, a number of experimental data are necessary. In this work, the retention data are predicted based on a single data point for a given solute/surfactant combination. The prediction of micelle/water partition coefficients with the COSMO-RS model was evaluated for the nonionic surfactant Brij 35 and the cationic CTAB for 21 solutes. The predicted partition coefficients along with the retention data of a single measurement were combined with a common retention model for the evaluation of partition coefficients. Thus, the description of the retention data based on a minimum of experimental data was realized in this work. The COSMO-RS model appears to be suitable for the flexible and qualitative prediction of the retention data, which is of special benefit, when a large number of solutes are studied. Furthermore, the introduced method is a promising alternative to describe the retention behavior of ionizable solutes, which is of great interest for e.g. pharmaceutical processes and analyses. To evaluate the method, lipophilicity profiles of several solutes were determined with MLC. While for the nonionic surfactant the lipophilicity is similar to the octanol/water system, a contrary phenomenon, namely a significant increase of the partition coefficient with progressive dissociation can be observed with the ionic surfactant. The MLC method proved to be an efficient method for the determination of micelle/water partition coefficients, excluding non-binding and "overbinding" solutes. The introduced COSMO-RS based approach for the estimation of the retention in MLC is especially promising for highly retarding solutes, since it is time and cost saving compared to the experimental determination.