Ten different mathematical models representing the electrophoretic mobility of analytes in capillary electrophoresis in mixed solvents of different composition have been compared using 32 experimental data sets. The solvents are binary mixtures of water-methanol, water-ethanol and methanol-ethanol, respectively. Mean percentage deviation (MPD), overall MPD (OMPD) and individual percentage deviation (IPD), have been considered as comparison criteria. The results showed that a reorganized solution model, namely the combined nearly ideal binary solvent/Redlich-Kister equation, is the most accurate model among other similar models concerning, both correlation ability and prediction capability.