Abstract: Plant breeding has played a key role in increasing agricultural productivity and meeting the increasing needs of the world, while the prevalence of genotype-by-environment interaction (GE) in multi-year, multi-location variety trials impedes variety selection and application efficiency. The Northwest Inland Cotton Production Region (NICPR) is currently the most important cotton-growing region, occupying more than 80% of the total cotton acreage in China. Therefore, mega-environment (ME) investigation and test environment evaluation are beneficial for the rational utilization of experimental resources and the improvement of the efficiency of cotton variety trials conducted in the NICPR. The objective of the present study was to demonstrate the application efficiency of the existing genotype main effect plus GE (GGE) biplot and a newly proposed location grouping (LG) biplot in exploring ME and comprehensively evaluating test environments using identification ability, representativeness, and desirability index based on the lint cotton yield of national cotton variety trials in the NICPR from 2011 to 2020. (1) The LG biplot revealed that the majority of test environments, including Shawan, Wujiaqu, Kuytun, Shihezi, Dunhuang, Bole and Jinghe, belonged to the same ME and suitably represented the targeting early-maturing cotton production region, while the test environment Usu was delineated out as an outlier of the early-maturing cotton ME. The test environment Makit in the medium-early maturing ME was also identified as an outlier in the southern Xinjiang cotton growing region, while the other test environments covering Bazhou, Alaer, Shache, Kuqa, Baicheng, Korla, and Tumxuk were all positively correlated and suitably represented the medium-early maturing ME. (2) The differences in identification ability among all test environments were not significant at P>5% level. The representativeness and desirability of Usu and Markit displayed significant differences from other test locations in the same ME, while the differences among other test locations were not significant. (3) According to the desirability index, the comprehensive ordination of the test environments in the early maturing cotton region was ranked as Shawan > Jinghe > Wujiaqu > Dunhuang > Bole > Shihezi > Kuytun > Usu. Similarly, on the basis of the desirability index, the ordination of test locations in the medium-early maturing cotton region was listed as Bazhou > Tumxuk > Alaer > Korla > Shache > Baicheng > Kuqa > Makit. It was clear that Usu and Makit should be removed from cotton variety trial scheme optimization for test efficiency improvement. The results of the study not only presented the highly efficient function of LG and GGE biplots in test environment evaluation in cotton variety trials in the NICPR and provided a theoretical basis for the optimization of the cotton regional trial schemes in Northwest Inland, but they also set a good example for future application in similar studies on other crops for other target crop growing regions.