This paper mainly discussed the spatial variation and distribution characteristics of soil pH in a mollisol watershed of northeast China by combining methods of classical statistics, with the Moran index model of global spatial autocorrelation and local indicators of spatial association and “3S” technology. The results showed that the mean value of soil pH was 6.72, the variation coefficient was 9.91%, and the variation range was 1000 m, which had strong spatial autocorrelation, the spatial variability was influenced significantly by structural factors such as terrain and parent material, and it was relatively severe in the east–west direction (E–W). Soil pH was significantly correlated the terrain factors such as the altitude and so on. Geographical weighted kriging showed that high value of pH mainly distributed in dry field in the middle while the low value area distributed in the eastern woodland and the western paddy field, presenting a trend as high in the middle and low in the east and west, which conformed with the distribution pattern of paddy land–dry land–forest from west to east. The results of global autocorrelation analysis showed a strong positive autocorrelation characteristic of soil pH and obvious differences in different directions. With the increase of distance, soil pH showed a negative autocorrelation characteristic, presenting isolation in space. The local spatial autocorrelation mainly displayed an H–H (high–high correlation) and L–L (low–low correlation) types, indicating a coexistence pattern of high and low value agglomeration. The H–H type mainly distributed in the middle flat place, where surface soil moisture and its spatial autocorrelation were both high. The L–L type mainly distributed in the eastern mountain and western paddy field, where surface soil pH content was low.