Soil microbial activity plays a key role in processes that affect ecosystem functions. Disturbances by management practices may change microbial biomass and activity. Inherited or due to management practices, soil microbial properties are spatially variable. The objective of this study was to reveal the heterogeneity of soil microbial properties in relation to other soil properties and management practices, and to determine microbial responses to grazing and cultivation in prairie soils.. Soil samples were taken from rolling upland mixed prairie of SW Oklahoma that were undisturbed or cultivated. The Undisturbed included ungrazed and different grazing at high or moderate intensities. Cultivated included continuous cultivation or once cultivated but being returned to pairie for at least 30 yrs (abandoned). Continuous cultivation included wheat, No-till cotton with rye as an intercrop, and conventional till cotton. Samples were taken along a 2 m transect with 20 cm intervals from each of the seven treatments. Based on coefficients of variation and information revealed by variograms on the spatial dependence between neighboring samples, dissolved organic C, microbial C, and microbial N were the most spatially variable of the measured soil properties. Although cultivated soils looked more uniform among the management systems evaluated at the landscape scale, these soils were the most variable in microbial biomass C and N. Dissolved organic C varied most in the grazed systems. Fitted variogram models for microbial properties showed that uncultivated soils had the strongest spatial dependence, and at smaller range. In the cultivated soils there was a periodic variation in the semivariance of microbial biomass C, organic C, and dissolved organic C (also termed the “hole effect”). Apparently, residue incorporation and root decomposition created favorable conditions for microbial growth in microhabitats that led to formation of microbial clusters where there is plenty of substrate. The assumption of most variogram models in geostatistics is that they are monotonic, i.e., semivariance increases with separation distance between samples. We find that such models are inadequate to describe the spatial dependence of soil properties in cultivated fields. In addition, the semivariance of dehydrogenase activity suggested a lower nugget effect and stronger spatial dependence than for microbial biomass C. Overall, microbial activities were greater in undisturbed, abandoned, or grazed systems when compared with those in cultivated ones, indicating partial restoration of the soil ecosystem. However, negative impact of cultivation on soil microbial community in abandoned soils was detectable, suggesting that 30 years of preservation was not long enough to erase human impact on a soil ecosystem.