To screen high quality cotton varieties with better suitability for production purposes.. The 24 early and early-mid maturing upland cotton varieties were used as materials for a 3-years comparative variety trials, and analyzed by correlation, principal component, clustering, gray correlation, etc. The coefficient of variation (CEV) of the 3 years test ranged from 0.40% to 4.18%, and the coefficient of variation was relatively high in 2022., between 1.01% and 20.48%, The greatest variation was observed in the number of bolls per plant and single boll weight; 6 pairs of correlations between traits reached the significant level, 14 pairs reached the highly significant level, the correlation between agronomic traits and yield and fiber quality traits was intricate and complex, and the internal correlations among fiber quality traits are relatively close; 5 principal components were screened, explaining 79.87% of the variability; and 5 major groups were obtained by clustering according to the varietal characteristics. Based on the affiliation function, gray correlation model and AHP model, a set of comprehensive resource evaluation system was proposed, and Jinfeng 6, J8031, Xinnongda Cotton 1 and other varieties with better comprehensive quality were screened. The comprehensive evaluation system was verified in resource groups and superior lines respectively, and the evaluation results were consistent with the field performance, which proved that the evaluation system has the basis for further popularization.