Although it has been suggested that several mechanisms can describe the direct binding of As(III) to organic matter (OM), more recently, the thiol functional group of humic acid (HA) was shown to be an important potential binding site for As(III). Isotherm experiments on As(III) sorption to HAs, that have either been grafted with thiol or not, were thus conducted to investigate the preferential As(III) binding sites. There was a low level of binding of As(III) to HA, which was strongly dependent on the abundance of the thiols. Experimental datasets were used to develop a new model (the modified PHREEQC-Model VI), which defines HA as a group of discrete carboxylic, phenolic and thiol sites. Protonation/deprotonation constants were determined for each group of sites (pKA=4.28±0.03; ΔpKA=2.13±0.10; pKB=7.11±0.26; ΔpKB=3.52±0.49; pKS=5.82±0.052; ΔpKS=6.12±0.12 for the carboxylic, phenolic and thiols sites, respectively) from HAs that were either grafted with thiol or not. The pKS value corresponds to that of single thiol-containing organic ligands. Two binding models were tested: the Mono model, which considered that As(III) is bound to the HA thiol site as monodentate complexes, and the Tri model, which considered that As(III) is bound as tridentate complexes. A simulation of the available literature datasets was used to validate the Mono model, with logKMS=2.91±0.04, i.e. the monodentate hypothesis. This study highlighted the importance of thiol groups in OM reactivity and, notably, determined the As(III) concentration bound to OM (considering that Fe is lacking or at least negligible) and was used to develop a model that is able to determine the As(III) concentrations bound to OM.