Sorption is one of the key process that affects the fate and mobility of pharmaceuticals in the soil environment. Several models have been developed for estimating the sorption of organic chemicals, including ionisable compounds, in soil. However, the applicability of these models to pharmaceuticals has not been extensively tested. In this study, we generated a high-quality dataset on the sorption of twenty-one pharmaceuticals in different soil types and used these data to evaluate existing models and to develop new improved models. Sorption coefficients (Kd) of the pharmaceuticals ranged from 0.2 to 1249.2 L/kg. Existing models were unable to adequately estimate the measured sorption data. Using the data, new models were developed, incorporating molecular and soil descriptors, that outperformed the published models when evaluated against external data sets. While there is a need for further evaluation of these new models against broader sorption datasets obtained at environmentally relevant concentrations, in the future they could be highly useful in supporting environmental risk assessment and prioritization efforts for pharmaceutical ingredients.