Calibration/validation experiments are critical to developing and testing soil moisture retrieval algorithms from microwave remote sensing platforms. Ground sampling must be optimally designed to minimize sources of error or bias during data collection. An important consideration in the design of ground sampling campaigns is that the in situ measured soil moisture is representative of the retrieval depth of a microwave sensor. Because of the popularity of portable impedance probe instruments, volumetric soil moisture is usually measured at a constant depth of approximately 0 to 6 cm and aggregated over a spatial extent representative of a pixel footprint. However, the retrieval depth of microwave signals can vary, and in agricultural soils is typically representative of the top ∼5 cm (or less) of the soil surface. It is unknown whether this mismatch between the depth of ground measured soil moisture and microwave retrieved soil moisture significantly contributes to error during sensor calibration/validation. This issue is addressed based on an analysis of over 3400 impedance probe measurements collected in 72 agricultural fields over four sampling occasions. Volumetric soil moisture was measured at depths of approximately 6 and 3 cm, at 24 locations in each field. The observed soil moisture sampling distributions and statistical moments are compared between these two measurement depths. Results demonstrate that the majority of fields exhibit a near-normal sampling distribution at both the 6- (60 of 72) and 3-cm (59 of 72) depths. An average difference of 0.016 m3 m−3 exists between all field means, generally indicating that wetter measurements are observed from the samples at 6 cm than those at 3 cm. Statistically significant differences were found between the measurement depths for field means (23 of 72), distributions (21 of 72), and variances (9 of 72). These findings demonstrate the importance of recognizing near-surface volumetric soil moisture spatial and depth variability effects during ground sampling experiments to minimize error in calibration/validation of retrieval algorithms.