For decades, forest ecologists and other researchers have attempted to relate forest growth (e.g., site index) or productivity (total biomass) to measurable soil water and nutrient characteristics. The attempts have not always been completely successful. Forest community and soil data collected from stands on national forest land in Illinois, Kentucky and Tennessee, provide insight into the problem of quantifying soil water and nutrient levels. Statistically, site productivity or biomass, as indicated by the basal area of old-growth stands, is not related to either the texture fractions (sand, silt, or clay) or water holding capacity (%) of either the A or B horizons. However, percent available water between field capacity and permanent wilting point (a function of texture), bulk density, and horizon depth was integrated to estimate the available water capacity (cm) for a horizon. Reducing this amount by the percentage of stone (particles > 2.0 mm) in the horizon, and summing the amount for all horizons to 90 cm (maximum rooting) or to a horizon impermeable to roots (bedrock or fragipan) produced a value that closely approximated the available soil water reservoir. When PAWC was transformed to a natural log scale, the relationship became more linear (r = 0.94; p < 0.0001). A strongly similar situation was found at LBL with regard to the relationship between white oak site index and calcium (lnCa; Kg/ha; r = 0.59, p <0.0001) and AWC (cm; r = 0.49, p < 0.0001) calculated for the rooting zone. In both regions, forest response was not statistically related to surface samples alone, thus leading to the conclusion that data from limited surface sampling is inadequate to identify forest-soil relationships.