The solubility of seven pharmaceutical compounds (paracetamol, benzoic acid, 4-aminobenzoic acid, salicylic acid, ibuprofen, naproxen and temazepam) in pure and mixed solvents as a function of temperature is calculated with SciPharma, a semi-empirical approach based on PC-SAFT, and the NRTL-SAC model. To conduct a fair comparison between the approaches, the parameters of the compounds were regressed against the same solubility data, chosen to account for hydrophilic, polar and hydrophobic interactions. Only these solubility data were used by both models for predicting solubility in other pure and mixed solvents for which experimental data were available for comparison. A total of 386 pure solvent data points were used for the comparison comprising one or more temperatures per solvent. SciPharma is found to be more accurate than NRTL-SAC on the pure solvent data used especially in the description of the temperature dependence. This is due to the appropriate parameterization of the pharmaceuticals and the temperature-dependent description of the activity coefficient in PC-SAFT. The solubility in mixed solvents is predicted satisfactorily with SciPharma. NRTL-SAC tends to overestimate the solubility in aqueous solutions of alcohols or shows invariable solubility with composition in other cases.