Multisensor capacitance probes (MCPs) have been used in many soil water-related fields. The manufacturer recommends a site-specific calibration before MCP use, and the calibration protocol requires replicated measurements of soil water content and MCP readings in the same soil volumes. In field research, such calibration is hardly plausible, and results cannot be extrapolated to plot average water contents in heterogeneous soils. A site-specific correction of the manufacturer calibration is a practical alternative to the field calibration in this case. The Typic Hapludult soil at the OPE3 USDA-ARS research site in Beltsville, MD was sampled in triplicate at the distance of 50 cm from four MCPs on three dates with distinctly different water contents. Both systematic and random differences between MCP and plot-average gravimetrically determined water contents were encountered. The manufacturer calibration led to the overestimation of low water contents and to the underestimation of high water contents. The depth-specific linear transformation of the factory calibration improved the estimation of plot-average water contents at all observation depths. Correcting MCP measurements for depth resulted in up to a 14.6% decrease in root-mean-square difference between MCP and plot-average measurements. Site-specific calibration correction may be useful when using MCPs in soil water monitoring.