Several tree height–diameter scaling rules have been proposed based on mechanical considerations or metabolic scaling. I used data on 202,950 trees of 86 species from the northeastern United States to examine height–diameter scaling patterns, and to determine whether they were sensitive to phylogeny (angiosperm vs. gymnosperm), shade-, drought-, and waterlogging tolerance, evergreenness, and wood density. Mixed effects models were used to estimate average scaling relationships and to quantify variability due to species and site attributes. Results did not support the invariant 2/3 scaling exponent assumed by the Metabolic Ecology Model. Evergreenness and wood density emerged as the primary predictor of height–diameter scaling in this assemblage, with considerable variation remaining at the species and location level. The results suggest that greater ecological and ontogenetic variability must be incorporated into scaling theories if height–diameter scaling is to be predicted successfully.