Zoige County, China, represents a fragile sub-alpine rangeland eco-environment with a severe land desertification problem. This paper aims at detecting land desertification change in Zoige County over 15 years with quantitative remote-sensing techniques using multi-spectral imagery. Landsat images acquired in 1994 and 2009 were analyzed using the following methodology: (1) image pre-processing; (2) spectral mixture analysis (SMA) to obtain precise sub-pixel classification results of land cover; and (3) change vector analysis (CVA) to conduct a multi-temporal comparison process. Change detection results depict the land desertification conditions and vegetation re-growth conditions. In this way, we characterized the spatial-temporal change pattern of land desertification in Zoige County between 1994 and 2009. After categorizing ecological regions based on change detection results, we analyzed the driving factors of both land desertification conditions and vegetation re-growth conditions, finding out that grasslands under intense grazing pressure tend to suffer severe desertification, while topographic relief has an obvious influence on vegetation re-growth. Specific suggestions for each ecological region are proposed, which can assist the development of environmental restoration measures and environmental protection measures in Zoige County in an effective way. Furthermore, this methodology for monitoring land desertification could be carried out across neighboring counties or in other regions with similar sub-alpine rangeland and land desertification problems.