Marine pollution has been considered an increasing problem because of the increase in sea transportation day by day. Therefore, a large volume of bilge water which contains petroleum, oil and hydrocarbons in high concentrations is generated from all types of ships. In this study, treatment of bilge water by electrocoagulation/electroflotation and nanofiltration integrated process is investigated as a function of voltage, time, and initial pH with aluminum electrode as both anode and cathode. Moreover, a commercial NF270 flat-sheet membrane was also used for further purification. Box–Behnken design combined with response surface methodology was used to study the response pattern and determine the optimum conditions for maximum chemical oxygen demand (COD) removal and minimum metal ion contents of bilge water. Three independent variables, namely voltage (5–15 V), initial pH (4.5–8.0) and time (30–90 min) were transformed to coded values. The COD removal percent, UV absorbance at 254 nm, pH value (after treatment), and concentration of metal ions (Ti, As, Cu, Cr, Zn, Sr, Mo) were obtained as responses. Analysis of variance results showed that all the models were significant except for Zn (P > 0.05), because the calculated F values for these models were less than the critical F value for the considered probability (P = 0.05). The obtained R2 and Radj2 values signified the correlation between the experimental data and predicted responses: except for the model of Zn concentration after treatment, the high R2 values showed the goodness of fit of the model. While the increase in the applied voltage showed negative effects, the increases in time and pH showed a positive effect on COD removal efficiency; also the most effective linear term was found as time. A positive sign of the interactive coefficients of the voltage–time and pH–time systems indicated synergistic effect on COD removal efficiency, whereas interaction between voltage and pH showed an antagonistic effect.