The thrips quantity ratio(TQR) model is an important tool for evaluating crop resistance to thrips based on the correlation between thrips quantities and cultivars. Unfortunately, it is inaccurate, and the results appear significantly inconsistent when analysing the same cultivars in the same field study. To improve this model, we first studied the resistance of 28 alfalfa cultivars to thrips in Cangzhou, Hebei Province, north China. The results showed that the most suitable evaluation period was from May to June, as the thrips population was stable during this period. Second, we found that the natural enemy population was significantly positively correlated with the thrips population density(R=0.7275, P<0.0001), which might influence resistance estimation. Hence, we introduced a parameter ‘α’, corresponding to the natural enemy quantity ratio, to eliminate the effect of the natural enemy using “αTQR”. Using the improved method, 28 cultivars were clustered into three classes: the resistant class, sensitive class, and median class. All numerical values were calculated for αTQR displayed as a Gaussian distribution. This information showed that all data should be divided into nine groups using a median value of 1±0.1 with an equal difference of 0.1. Based on the new standard cultivars, Gongnong 1, Alfaking, Cangzhou and Algonquin were classified as highly resistant cultivars; Zhongmu 3, Gongnong 2, Zhongmu 1 and Zhongmu 2 were classified in the resistant group; Queen was classified in the moderately resistant group; Derby, WL354HQ, KRIMA, Apex, 53 HR, SARDI 5 and Farmers Treasure were classified in the median class; WL319HQ, WL343HQ and Sitel were classified as the low sensitive group; WL440 HQ and SARDI7 as the moderately sensitive group; WL168HQ and Sanditi as the sensitive group; and SARDI 10, WL363HQ, FD4, WL323 and SOCA as the highly sensitive group. The thrips quantity ratio (TQR) model is an important tool for evaluating crop resistance to thrips based on the correlation between thrips quantities and cultivars. Unfortunately, it is inaccurate, and the results appear significant inconsistent when analyzing the same cultivars in the same field study To improve this model, we first studied the resistance of 28 alfalfa cultivars to thrips in Cangzhou, Hebei Province, north China. The results showed that the most suitable evaluation period was from May to June, as the thrips population was stable during this period . Second, we found that the natural enemy population was significantly positively correlated with the thrips population density (R = 0.7275, P <0.0001), which might influence resistance estimation. Hence, we introduced a parameter ’α’, corresponding to the natural enemy quantity ratio, to eliminate the effect of the natural enemy using “αTQR”. Using the improved method, 28 cultivars were clustered into three classes: the All numerical values ​​were calculated for αTQR displayed as a Gaussian distribution. This information showed that all data should be divided into nine groups using a median value of 1 ± 0.1 with an equal difference of 0.1. Based on the new standard cultivars, Gongnong 1, Alfaking, Cangzhou and Algonquin were classified as highly resistant cultivars; Zhongmu 3, Gongnong 2, Zhongmu 1 and Zhongmu 2 were classified in the resistant group; Queen was classified in the moderately resistant group; Derby , WL354HQ, KRIMA, Apex, 53 HR, SARDI 5 and Farmers Treasure were classified in the median class; WL319HQ, WL343HQ and Sitel were classified as the low sensitive group; WL440 HQ and SARDI7 as the moderately sensitive group; WL168HQ and Sanditi as the sensitive group; and SARDI 10, WL363HQ, FD4, WL323 and SOCA as the highly sensitive group.