In this research, the photocatalytic process is used for the treatment of spent caustic wastewater from petrochemical industries. For this purpose, by use of two types of synthetic photocatalyst (synthetic zinc oxide (ZnO-Syn) and combined (composite) zinc oxide with Fe (ZnO-Fe3O4)) in a photoreactor, and measuring removal percentage of chemical oxygen demand (COD), results are modeled with the design of experiment (DOE) and artificial neural network (ANN) methods. According to the implemented calculations, it can be concluded that the ANN is a more suitable method than the DOE in modeling and forecasting the amount of COD. Modeling of this research showed that increasing the concentration of ZnO-Fe3O4 and ZnO-Syn photocatalyst in a state of neutral pH, in optimal amount of 1.08 and 1.29 g/L, leads to enhance the COD removal up to 88% and 74% without restrictions, respectively, and also 2 g/L for both of them with restrictions leads to 80% and 69% removal efficiency, respectively. In addition, the study of the parameters' effects, including oxidizer amount, aeration rate, pH and the amount of loaded catalyst, indicate that all factors except pH have had positive effect on the model. Also, photocatalyst acidic pH is more suitable at low concentrations of the photocatalyst. Besides that, by increasing pH, the efficiency of removal will be reduced when oxidant is at its low level. The results showed that photolysis and adsorption adoptions have a very small effect on efficiency of COD removal compared with the photocatalyst adoptions and it is negligible. In addition, the photocatalytic method has an acceptable capability for removing phenol in wastewater samples, whereas it is inefficient for reducing the sulfide solution in wastewater.