A predictive approach seems useful to study human and veterinary pharmaceuticals in the environment and provide an idea of overall levels of contamination, so as to restrict monitoring to those molecules which are most likely to represent possible environmental contaminants. Predicted environmental concentrations (PECs) can be calculated by a mass balance approach, while a recent proposal from the European Agency for the Evaluation of Medicinal Products (EMEA) suggests an alternative method for calculating PEC for each pharmaceutical and then focusing further work on molecules with high PEC values. We used the results of monitoring campaigns on the River Po, in Northern Italy, to assess the accuracy of predictive models with measured environmental concentrations (MECs). The comparison indicated that in some cases a refined PEC value can provide a good approximation of the MEC. In other cases PECs substantially differed from the MECs, particularly when there were not enough data to estimate the environmental fate of the molecule. Predictive models might therefore be useful for studying pharmaceuticals in the environment, providing enough experimental data is available on the environmental fate of the molecules.