The present work examined the removal of methylene blue (MB) from aqueous solutions through an electrocoagulation/hazelnut shell (EC/HNS) adsorption coupling process. The process was evaluated using a Box–Behnken design (BBD) under the response surface methodology (RSM). According to the obtained results from BBD–RSM regression analysis, we found that the experimental data are best fitted to the second-order polynomial model with coefficient of determination (R²) value of 0.9534, adjust coefficient (R_adj²) value of 0.9926 and predicted correlation coefficient (RP_red²) value of 0.9973. The maximum MB removal efficiency was obtained under optimized conditions (HNS dosage, 6 g/L; reaction time, 17.75 min; and current density, 2.6 mA/cm²). Under the optimum conditions, the MB removal rate reached 99.45%, and the RSM predictive value was 98.71% with a small deviation. These results confirmed that the RSM model accurately and reliably analysed the process of removing MB from wastewater.