Response surface methodology (RSM) and artificial neural networks (ANNs) based on a multivariate central composite design (CCD) were applied to model and optimize the photocatalytic degradation of N,N-diethyl-m-toluamide (DEET). The individual and interaction effects of three main operating factors (mass of TiO2, initial DEET concentration, and irradiation intensity) on process efficiency were estimated, proving their important effect on % DEET removal. Among the independent variables, TiO2 concentration displayed the highest effect on DEET degradation followed by initial DEET concentration and UV intensity. The optimization and prediction capabilities of ANNs and RSM were compared on the basis of root mean squared error, mean absolute error, absolute average deviation, and correlation coefficient values. Results proved the usefulness and capability of the experimental design strategy for successful investigation and modeling of the photocatalytic process. Moreover, the selected ANN gave better estimation capabilities throughout the range of variables than RSM. Based on the models and the related experimental conditions, the optimal values of each parameter were determined. Under optimum conditions, DEET and total organic carbon (TOC) followed pseudo-first order kinetics. Nearly complete degradation of DEET took place within 15 min whereas high TOC removal percentages (>85%) was achieved after 90 min irradiation time.