The protein and amino acid composition of 24 rice samples from 8 late indica varieties in three production bases were determined. With FAO/WHO provisional pattern of amino acids as reference, a comprehensive evaluation model for rice protein was established by combining modified analytic hierarchy process (AHP) with grey correlation analysis, and the feasibility of the evaluation model in evaluating the nutritional value of protein was verified. The results showed that the quality of rice protein was significantly different among different bases and varieties (P<0.05). The correlation coefficient r of the weighted grey correlation method with PER and NPR was 0.959 (P=0.003) and 0.974 (P=0.001) respectively. The results demonstrate that the weighted grey correlation degree is highly correlated with biological indexes of protein quality, which can characterize the nutritional quality of rice protein scientifically and effectively.