This experiment was conducted from April to September in 2009 in Baotou-lake Farm of Korla city, Xinjiang region, China. Two experiment schemes were designed for analyzing the spatial variability and lateral orientation of monitoring points for soil moisture. In scheme 1, 44 monitoring points were positioned with sampling depth of 0~60cm. In scheme 2, 20 monitoring points were placed with sampling depth of 0~80cm. Samples were taken in the whole cotton growing period before and after irrigation in both schemes. Statistical analyses including Q-Q test, descriptive statistics, t-test and geo-statistical analysis were performed to the soil moisture data. The results indicate that the increase of soil moisture can enhance their spatial variability. In drip-irrigated cotton cultivation, spatial variability of soil moisture resulted from the combination of several random variables and the structural factors such as climate, topography, soil form. Semi-variogram models belong to spherical model and the range of spherical model for soil moisture is about 9.40m~35.35m. The accuracy of the fitted model decreases as the soil moisture content increases. The monitoring points should be placed at the range of 0~0.475m from the drip tape and it is not suitable to place them at a farther place. It can be concluded that it is best to position the monitoring points at outward cotton row in consideration of management and soil moisture monitoring. The outcome of this study can be used for the proper design and placement of monitoring points for soil moisture.