The performance of different activated carbons for the removal of nonionic surfactants from metallurgic wastewater was assessed through lab scale experiments. Two different matrices were used: a simple one, obtained by dissolving different amounts of a single nonionic surfactant (Triton X100) in distilled water, and a complex matrix, prepared as above but using surfactants-free wastewater from an Italian metallurgic plant as solvent. In this way the main operative parameters that affect the adsorption process in the simple matrix have been studied avoiding the interference due to the complex matrix: then the results were utilized to define and optimize the tests carried out on the complex matrix. The adsorption equilibrium experimental data were best fitted with a Langmuir isotherm, allowing defining the contact time and the proper design parameters for the adsorption column. The different tests were performed on four different activated carbon types, and the removal efficiency and the treatment cycle duration for each of the tested carbons were discussed and compared. The experimental results showed that the saturation adsorption capacity is not notably correlated either with the pH value or with the water matrix, whereas the slope of the isotherm, defined by the Henry constant, is sensibly higher at strong acidic or alcaline conditions, with a minimum value at nearly neutral pH. Therefore, it was concluded that the removal efficiency is maximized when the operative pH was in the 2-4-unit value range. The best activated carbon, in terms of removal efficiency, resulted to be a mineral activated carbon, characterized by the highest iodine number, and thus with the largest porosity. Removal efficiencies were in the 60 to 80% range.