Abstract: The soil samples in the 44 km2 range around a tailing bank were sampled. The spatial autocorrelation analysis and geostatistics method were used to study the spatial autocorrelation and variability of uranium in the soil around uranium tailing bank. The results of geostatistics method show that the mean value of uranium content in soil is 2.9 times of the local background value. According to the single-factor pollution index, the proportion of clean samples in the samples taken is 0.5%, that of mild pollution is 22.16%, that of moderate pollution is 46.76%, and that of heavy pollution is 31.03%. According to the spatial autocorrelation analysis, the global Moran’s I of the uranium is 0.154 3, the Z value is 2.066 4, and the significance is greater than 1.96, indicating that the uranium in the soil around the uranium tailing bank has positive spatial autocorrelation and spatial aggregation. The local indicators of spatial association (LISA) aggregation map of uranium were obtained by means of local autocorrelation analysis, and the correlations between the properties of each unit and similar units in the area were discussed. The results of the semi-variance show that uranium is strongly correlated with the results of spatial autocorrelation analysis. According to the results of local autocorrelation analysis, the study area is divided into no-till farming area, comprehensive management area, key control area and key protection area. And it is suggested that the concerned departments carry out zoning management according to different pollution degrees.