To improve the efficiency of protein extraction from broken long-grain rice, a quadratic rotatable central composite response surface methodology (qrCCD-RSM) was utilized to investigate the affecting mechanism of alkali concentrations, broken long-grain rice particle sizes and solid-liquid ratios (SLR), based on the results of Min-Run Res IV factorial design and steep accent experiments (SAE). The optimal extraction condition was obtained, i.e. 80 mesh of broken long-grain rice was used to extrat proteins, soluting in 0.09 mol/L of NaOH buffer, with the SLR 1 g13.5 mL at 50 ℃ for 120 min. Under this optimized condition, the protein extraction rate and its purity was found to be (82.25±2.53)% and (91.68±1.26)%, respectively.