To reveal the relationship between the SPAD value and its derived value of rice leaves and the protein content of rice in the cold regions, this study measured the top three leaves at the jointing stage (T1), booting stage (T2) and heading stage (T3) of rice in 2020 and 2021. Based on the data of the pot experiments (Experiment 1 and Experiment 2), the relationship model between the SPAD\|derived index and the protein content of rice was established, and the data of the field experiment (Experiment 3) was used to verify the established model. The results showed that in the nitrogen fertilizer experiment in 2020, the protein content of rice at A8 level was significantly increased by 34.55%, 27.44%, 26.39%, 22.19%, 18.07%, 14.39%, and 12.23% compared with A1 to A7 levels, respectively. The taste value of A8 level was higher than that of A1~A7 was significantly reduced by 8.10%, 5.06%, 4.99%, 4.10%, 3.45%, 2.96%, and 2.28%, respectively. The changes in protein content and taste value in 2021 were the same as the former. Compared with C5, C4, C3, C2, and C1, the protein content of C6 was significantly increased by 2.99%, 12.23%, 10.43%, 5.04%, and 15.63%, respectively, and the taste value of C6 was lower than that of C5, C4, C3, C2, and C1 and were significantly reduced by 1.17%, 12.09%, 3.54%, 2.89%, and 7.93%, respectively. Variety differences and nitrogen application rates had a greater impact on the SPAD distribution of the top three leaves of rice at different growth stages, but the two “black and yellow alternation” phenomenon in rice canopy leaves were not affected by the variety, among which the single leaf SPAD value was affected by varietal differences and had no correlation with protein content. With the help of indexes SPAD(L1+L2+L3)/3, SPADL2×L3/mean, the effect of cultivar and environmental differences on the prediction of protein content was effectively reduced. Influence, the fitting equations of index SPAD(L1+L2+L3)/3 and protein content in T1~T3 period are: Y=0.24X-1.94,R2=0.95**, Y=0.25X-1.69,R2=0.94**,Y=0.27X-2.45,R2=0.92**; SPADL2×L3/mean fitting equations were: Y=0.22X-1.05,R2=0.95**,Y=0.27X-2.43,R2=0.92**,Y=0.26X-2.24,R2=0.92**, Y is the protein content, X is the SPAD derived value, and the fitting equations all reached the extremely significant level. The protein content of rice and the taste value score were linearly negatively correlated, the regression equation was Y=-4.21X+113.32 (Y is the taste value, X is the protein content), and the goodness of fit R2=0.93**, reaching a very significant level. To sum up, with the help of the indexes SPAD(L1+L2+L3)/3 and SPADL2×L3/mean, the purpose of fast, non\|destructive and real\|time detection of rice protein content was able to be achieved, and to a certain extent, the quality of cooking and eating quality of rice were determined, and the results could be achieved. According to the requirements of quality harvesting and real\|time monitoring of quality, the sustainable development of high\|quality cold rice was promoted. |