In the present study, nonlinear modeling of color and chemical oxygen demand (COD) removal of Basic Blue 3 as a fairly toxic textile dye by photocatalytic degradation using TiO 2 was investigated. A three replicates and one duplicate of Box-Behnken design with 45 runs, with three factors and three levels of catalyst concentration of 0.5-1.5 g/L, initial dye concentration of 20-100 mg/L, and initial pH of 3-11 were set to quantify the explanatory variables. ANOVA results showed that the best-fit multi-nonlinear regression models were cross-validated (R 2 pred ) accounting for 99.98-99.79 % and were expressed (R 2 adj ) accounting for 99.88-99.85 % of variation in decolorization (color removal) and mineralization (COD removal), respectively. In total, 0.51 g/L of TiO 2 , 100 mg/L of initial dye concentration, and initial pH of 7 were established as optimum operating parameters by leave-one-out optimization that resulted in 88.61 and 79.03 % for color and COD removal, respectively. 88.78 % of color removal and 76.95 % of COD removal were determined by cross-validation experiments at optimum conditions. Results showed that optimum point determination was successful.