The Northwest inland cotton growing regions (NICR) is currently the most important cotton producing area in China. The scientific classification and comprehensive evaluation of national cotton varieties over the years in this cotton area is beneficial to the rational utilization of national cotton varieties resources and the improvement of cotton production efficiency. According to the national registration bulletins of cotton cultivars, up to 37 cotton cultivars in this region were registered in the national level from 2003 to 2019. These cultivars were evaluated and classified using genotype by yield-trait (GYT) biplot analysis based on their levels in combine with lint cotton yield and other important characteristics including pre-frost yielding rate (PFR), fiber length, fiber strength, micronaire, fusarirum wilt index (FWI), and verticillium wilt index (VWI). Results showed that the 37 cotton cultivars can be clustered into three types with distinct trait profiles. Type I consisted of 17 cotton cultivars, namely Xinluzao 13, Zhongmiansuo 49, Xinluzao 21, Ba 13222, Xin 46, Tianyun 0769, Z1112, Xinshi K18, J206-5, Xinshi K21, Hemian A9-9, Chuangmian 508, H33-1-4, Jinke 20, Xin K28, Chuangmian 512, and J8031, showing high levels of combinations between yield and other target traits, and was the most suitable for cotton production in the region. Type II was characterized by poor performance in FWI, and moderate yielding ability and fiber quality traits, and was of limited application in this area. Type III was characterized by an excellent performance in FWI but poor performance on the other traits, and cultivars of this type may be useful for sources of disease resistance. Cultivars Chuangmian 512, J8031, Xin K28, H33-1-4, Jinke 20, Xinluzao 13, and Chuangmian 508 were selected for their high superiority index, such as high levels of combination between yield and other traits, while Xinlumian 1, Xinluzao 33, Chuangmian 501, and Xinluz 51 were inferior in this context. The GYT biplot analysis in this study differed from traditional index selection in that it was based on the combinatoin level of yield and other traits rather than on the levels of individual traits. In GYT biplot the various yield-trait combinaitons tended to be positievly correlated, allowing for easy visualizationand objective selection of cultivars based on multiple traits. Compared with the more traditional genotype by trait (GT) biplot, GYT biplot normally explained more of the total variation and achieve higher goodness of fit. In addition, the method used in this study could be applied to other regions and crops in this reasearch topic.