Abstract The use of detailed county soil surveys for predicting site productivity was evaluated for taxation purposes. One hundred and thirty permanent forest inventory plots containing suitable site index trees for loblolly pine (Pinus teada L.) were located on 11 different soil mapping units in the Sam Houston National Forest in Texas. The correlation between the field site indices for the mapping units and the site indexes projected by the Soil Conservation Service (SCS) was extremely low (r = .10). Analysis of variance for the soil mapping units suggested that the actual average site index between mapping units was significantly different. Duncan's multiple range test, however, showed that these differences were limited to those soils having extreme site index values, with the majority of the mapping units showing no significant difference in productivity. Detailed county soil surveys provided an inadequate basis for determination of productivity as defined by site index for taxation purposes.