The purpose of this study is to measure the variation in tree growth by individual soil taxonomic units on upland sites supporting natural forest stands, and to relate tree growth to characteristics of the soils. Four species occurring in natural, even-aged stands were chosen as indicator trees of the major forest types: jack pine (Jack Pine type), northern red oak (White Pine-Northern Red Oak-White Ash Type), sugar maple (Northern Hardwoods Type), and big tooth aspen (Aspen type). Growth measurements (site index and cubic foot volume) and soil descriptions for 138 field plots on twelve soil taxa were obtained on glacial parent materials of Cary and Valders age. Jack pine on seven soil taxa and northern red oak on four soil taxa exhibited distinct significant differences in growth among soil taxa. Sugar maple on five soil taxa showed no statistically significant differences in growth among the soil units. But, the variability on two of them was acceptable by the criteria used of site indices in this study. Sugar maple is not very responsive to wide variation in soil properties. Big tooth aspen exhibited statistically significant differences in growth among four of eight soil taxa, but on one soil soil unit was the variation in growth considered acceptable. The lack of differences for big tooth aspen is attributed to genetic variability of the species. Rather wide variations in growth on one or more soil taxa were observed for each species studied and these variations exceeded the assumed acceptable levels used for this study. Only a few soil taxa were found to consistently support better or poor growth for all species. Several soil properties, used for taxonomic purposes, which vary considerably within soil units were found related to variation in tree growth. These include depth to mottling, depth to finer textured lenses in sandy soils, thickness of A and B horizons, thickness of spodic horizon development, and texture of subsoils. Soil taxonomic units expressed as soil management units showed the general trend of least growth on the coarest textured soils. This was evident for all species. As the texture became finer, or soil horizons more strongly differentiated, the growth improved. Beyond a range characteristic for each species, the growth decreased with increasing finer textures. Several soil management units showed significantly more variation in productivity for all species of trees than when expressed as individual soil units. Several conclusions drawn from this study are: (1) a few soil taxa (Grayling, Montcalm and Graycalm, for example) can be used in evaluating site productivity for pine and oak forest types, but little reliance can be placed on any one soil taxa for evaluation of productivity for maple and aspen types; (2) some soil taxa (Rubicon, for example) must be refined by separating into smaller, less variable units, before they will yield useful estimates of forest productivity; (3) soil taxonomic units expressed as soil management units show a pattern of growth rates with differences in texture for jack pine, red oak, sugar maple and big tooth aspen.