Ecological stability, changes of eco-system under natural and human-induced disturbances such as fire, grazing, cultivation etc., involved in how and to what extent human can utilize natural resources but without inducing eco-system degradation or collapse, is a problem worthy of study. From the reaction of an eco-system to disturbances, it has been concluded that ecological stability could be defined as resistance and resilience. Because of many complicated factors such as the type, the intensity, the temporal scale and spatial scale of disturbances, it has been difficult to make qualified measurements for a real community. Therefore, researches related to resistance and resilience has been stopped at concepts, and studies on qualified resistance and resilience about a real community were scarce. In order to make an tentative example for study on this topic, 6 communities in different desertification stage were sampled in a sandy desertified region, and community investigation was done one time for every half month from the beginning to the end of growth season. With the investigation data made at the same time in continual two years before and during a special period when climate was extremely arid, an resistance index, which indicates the resistance stability of the 6 communities, were defined and calculated as the ratio of vegetation data at the same time in two years. It could be expressed as: resistance index = the vegetation data at the same time in the year during which the aridity event took place / the vegetation data in the year before aridity event took place. Results showed that, the resistance indexes of the other 5 communities calculated with living biomass coverage were between 0.1760 and 0.0217, decreasing with enhanced desertification except for community 4, of which the resistance index was higher with a value of 0.8615 because of Artemisia halodendron Turcz. ex Bess, an semi-shrub species which was dominated in the community. While the resistance indexes calculated with standing litter coverage and ground litter coverage were between 0.4394-2.4128 and 0.7983-5.3810 respectively, showing no consistent trend with the desertification level of the 6 communities, but be in line with the fact that standing litter and ground litter was less affected by drought and more stable in this extremely arid disturbance event. By further regression analysis made between living biomass coverage and each part of its components such as annual plants coverage and perennial plants coverage, showing that the correlationship between living biomass coverage and annual plants coverage was lower and weak, while that between living biomass coverage and perennial plants coverage was higher and significant, the source and constituents of the resistance could be understood and it could be concluded that living biomass was mainly consisted of perennial plants under this extremely aridity event. Therefore, it could be concluded that the resistance was mainly from perennial plants. Through the quantitative definition of vegetation stability with resistance, it could be concluded that, loss of biodiversity, especially loss of perennial plants, was the main mechanism and fundamental reasons for decreased community stability in desertification. The results could be explained from a reverse aspect as that biodiversity promoted the 6 desertified communities′ resistance stability under the short period of aridity disturbance, and also could be seen as an answer to the ecological questions such as “biodiversity′s function in ecological system” and “biodiversity and stability”.