Various thermodynamic models were tested concerning their applicability to predict gas-liquid solubilities, relevant for synthesis gas conversion to methanol, higher alcohols, and hydrocarbons via gas-slurry processes. Without any parameter optimization the group contribution equation of state (GCEOS) turns out to be the best model with an average, relative deviation of 19.0%. Ifa single binary interaction parameter is optimized for each binary system, the Peng-Robinson equation of state, the regular solutions theory, and the Flory-Staverman model all give good predictions with average, relative deviations of 4.0, 10.4, and 10.0%, respectively. As expected, the predictions from these models improve further and agree excellently with the experimental values by optimizing two binary interaction parameters for each binary system (average relative deviations < 2% for all models). The gas-liquid solubilities could also be correlated accurately to the temperature (average relative deviation = 2.1%) by assuming a constant enthalpy of solution (CEOS) model. For particular binary systems the Flory-Staverman model and the CEOS model give also reasonably accurate predictions of the gas-liquid solubilities by calculating the binary interaction parameters from pure component properties. Such an approach is promising for predicting as yet unknown gas-liquid solubilities without the need for experimental data.