Validation of ground-based microwave remote sensing measurements with point measurements of soil moisture remains an important task. The number of point measurements needed to characterize the mean soil moisture of an area on the order of tens of square meters, and the manner in which these observations should be made remain unknown. We conducted a field experiment in central Iowa to characterize the spatial process of soil moisture and sampling strategies for estimating the mean soil moisture of an area that represents the footprint of a ground-based microwave radiometer. We determined the spatial dependence of the moisture content of bare soil at this scale, which allows the determination of the mean soil moisture from point measurements. This spatial dependence can be described by an exponential variogram model. The parameters of this model vary according to mean moisture conditions. The accuracy of the estimate of mean soil moisture depends on the number of point measurements, what sampling technique is used, and mean moisture conditions. When we use spatial dependence, we increase the accuracy and precision of our estimates of the mean. Lattice and stratified sampling schemes require fewer measurements than random sampling when a higher level of accuracy in the estimate of mean soil moisture is desired. Practically, if measurements of soil moisture are made in random locations and spatial dependence is not used, we found that approximately 20 measurements within a 7 by 9 m footprint would estimate the mean to within close to 0.01 m3 m−3 of the true value for the entire range of soil moisture conditions that we observed.