Depth to subsurface features and diagnostic subsoil horizons is important when classifying and interpreting soils. Current techniques used to estimate these properties for map unit composition include linear transects. This two dimensional technique may be improved with a three dimensional approach by creating a subsurface model of the diagnostic horizon or feature. Accurate assessment and predictions of subsurface soil features will aid in resolving correlation discrepancies in multi-county maintenance and ultimately, soil surveys of major land resource areas. This study was conducted to evaluate the usefulness of real-time Differential Global Positioning System (DGPS), and Geographic Information System (GIS) software in determining and predicting depth to soil diagnostic subsurface features. A 200m by 200m area was selected in Hamilton County, Tennessee to study depth to lithic contact. Depth to hard sandstone bedrock governs most soil interpretations in the Cumberland Mountains and Plateau major land resource area (MLRA 125). Point data were gathered by walking across the study area in a grid-like fashion. Depth to bedrock was documented for each point using hand tools and locations were documented using real-time DGPS. Depth to hard bedrock models were constructed using four different interpolation methods available in ArcGIS^® Spatial Analyst. Models were evaluated using a separate set of random observation points to determine which method provided the highest accuracy. Of the methods evaluated, the Natural Neighbor method was found to have the lowest Root Mean Square Error (RMSE) and 95 percent accuracy range. This method was also preferred because it maintained the observed minimum and maximum values of this dataset. Further evaluation of the Kriging method is planned.