Abstract: Southern Xinjiang is one of the most important cotton-producing areas in China. It is necessary to fully understand the spatial distribution of cotton and variation characteristics for national grain and cotton supply, particularly on the development of the cotton textile industry in China. Therefore, this study followed the research idea of "reconstructing growth curve - extracting planting information - analyzing changing characteristics". Firstly, TIMESAT software was used to generate the enhanced vegetation index (EVI) growth curve of cotton in Southern Xinjiang. Subsequently, a Double-Logistic filter was selected to rebuild the growth curve. Secondly, the specific characteristics of the cotton growth curve were analyzed further to obtain the cotton growth threshold. Thirdly, a Band Math tool in ENVI5.3 was selected to extract the cotton planting areas. The spatial distribution accuracy of extracted datasets was then verified using Google Earth high-resolution image. Finally, a systematic analysis was made on the temporal and spatial variation characteristics of cotton planting from multiple perspectives. The results showed that: 1) The spatial distribution pattern of cotton was basically consistent with the soil and water conditions, where mainly distributed in the south of Tianshan Mountains and clustered in the northeast of southern Xinjiang, indicating a "core-edge" structure with Aksu region as the core, while Kashgar and Northern Bazhou as the margin. 2) There were significant differences between type I cultivated land and other types in different years, indicating the pretty obvious spatial differentiation. The active regions of cotton planting variation were mainly distributed in Aksu, Kashgar, and northern Bazhou, indicating the main cotton-growing regions in southern Xinjiang. There was the most significant correlation in the flow conversion between type I and type II cultivated land, grassland, and artificial land surface, indicating that the flow increased sharply. 3) The spatial distribution of cotton showed the "northeast to southwest" trend. The cotton planting center basically kept stable in Aksu City after a major migration in recent 20 years, with a total migration distance of 91.5 km and an annual migration rate of 4.58 km/a. 4) In detecting "hot spots" of cotton planting areas, the cold spots were mainly distributed in Kezhou and Hetian in southern Xinjiang, indicating a gradual concentration to the southwest after 2005. Correspondingly, the distribution pattern of hot spots changed significantly from year to year. Furthermore, the hot spots were mainly distributed in Aksu prefecture before 2005. The hot spots gradually extended to the northeast of southern Xinjiang after 2005, where mainly concentrated in Aksu prefecture and the north of Bazhou. Consequently, the temporal and spatial variation characteristics of cotton planting using EVI data can widely be expected for large-scale, long-term information monitoring. The yield estimation model can also be further constructed using the cotton growth curve, as well as the relationship with cotton actual output. Finally, quantitative remote sensing can be realized on cotton yield prediction. The findings can provide sound support to optimize the cotton structure distribution for the decision-making and formulation of regional land management.