Abstract A technically viable and alternative treatment option for the removal of arsenic from water is the constructed wetland (CW) technology. Modelling has emerged as a powerful tool for increasing understanding of water treatment systems such as CWs. The present contribution presents the first and second step implementation of the two most significant arsenic retention processes that take place in CWs (aqueous complexation followed by arsenic precipitation, and arsenic sorption in the granular media) in RetrasoCodeBright (RCB) (giving the RCB–ARSENIC model), showing the first simulation results (on the physical and chemical processes). The aqueous complexation reactions included in the RCB–ARSENIC model were considered equilibrium reactions. One sorption reaction and six mineral precipitation–dissolution reactions were included in the model. Five theoretical case studies (total arsenic concentrations from 23.5 to 300μg/L) were simulated in order to test the model resulting from the first step implementation. Once the second step had been implemented, the model was checked and then tested in a CW prototype operating without plants, considering different total arsenic concentrations at the inlet (from 50 to 300μg/L). In terms of total As concentrations in effluent, the simulated data closely matched the data measured in most evaluated cases. The iron and arsenic species relationships, as well as the arsenic retention percentages obtained from simulations, were in agreement with the literature. According to the quality of the obtained results, at this stage it can be considered that the first and second step implementation of the RCB–ARSENIC model has provided reasonably good response values.