the spatial and temporal variability of soil moisture is high due to the influence of several environmental features and the land-use. the understanding of this variability plays a fundamental role in the modeling of surface runoff, soil erosion and sediment transport. this study aimed to evaluate the spatial and temporal patterns of surface soil moisture over the annual seasons, in a watershed with predominance of oxisols in the south of the state of minas gerais. for this purpose, the temporal stability concept was used as well as semivariogram models to compare the spatial structure of the adjusted semivariogram models. it was possible to detect a strong spatial dependence of soil moisture in the watershed, with a spatial dependence degree of > 80 %. besides, spatial pattern of the scaled semivariograms was similar for summer and fall, and different for winter and spring. based on these results, the spatial patterns of soil moisture differed over the year, and were most homogenous in the rainy season (summer). differences in the temporal patterns of soil moisture according to the season were verified, and also, the sperman test demonstrated a temporal bias in soil moisture i winter and spring.