Display omitted •A resin concentration of 40 g·L-1 and a 50% v/v ethanol–water solution were found to be as the best options to recover phenolic compounds.•Pseudo-second order kinetic model best the experimental data, Langmuir isotherm modelled the adsorption process for the MN resins and Freundlich isotherm for the PAD resins.•The MN resins were more suitable, presenting higher percentages of adsorption/desorption.•It is possible to recover 90% of the initial phenolic compounds, with a low presence of sugars the MN200 resin. Agroindustry wastewater represents an opportunity to recover high added value antioxidants such as phenolic compounds. An adsorption/desorption process was investigated to recuperate these compounds using Purolite non-ionic resins (MN200, MN202, PAD900 and PAD950). The study was conducted with a model solution containing 1.1 g∙L-1 of tyrosol and 0.2 g∙L-1 of catechin, which are two of the main phenolic compounds found in olive mill wastewaters, and with a real olive mill wastewater. The main objective was to determine the optimal concentration of resin and the best desorption solvent for the maximum recovery of phenolic compounds. For it, the process kinetics were determined, and the adsorption mechanisms were characterized by means of isotherm models. Results showed that the pseudo-second order kinetic model fitted to the experimental data, while the Langmuir isotherm correctly modelled the adsorption process for the MN resins, whereas the Freundlich isotherm was the model that best described the adsorption process with PAD resins. A resin concentration of 40 g·L-1 and a 50% v/v ethanol–water solution were selected as the best options for recovering the phenolic compounds. The tests with olive oil washing wastewater showed that, for some of the resins, other compounds present, such as sugars, interfered with the adsorption of phenolic compounds, reducing the effectiveness of their recovery. Finally, the MN200 resin was selected as the best adsorbent. It achieved a recovery of 91% of the phenolic compounds present in the initial wastewater, finding only 5% of the sugars measured in the initial OOWW sample, in the final stream.