The Toxics Release Inventory (TRI) is the most comprehensive and widel
y reported information on hazardous discharges to the environment in t
he United States. Unfortunately, the fledgling nature of the TRI may l
ead to simplistic interpretations of the results. In particular, TRI s
ummaries typically report total releases of toxic chemicals by weight,
implicitly assuming that a discharge of substance A is equivalent to
an equal weight discharge of substance B. In contrast, various indices
of the toxicity to humans of the chemicals reported in the Inventory
indicate that the most harmful are more than 1,000,000 times more toxi
c than the least harmful. Thus, the simple rankings and time trends of
facilities, industries, counties, and states as sources of toxic rele
ases can be misleading because they have neglected relevant toxicologi
cal data. We contrast the one-to-one ranking of the TRI data with a ra
nking based on relative toxicity, using ''threshold limit value'' (TLV
) indices. The weighting scheme presented here is a useful first step
for correcting the TRI but is certainly not definitive. Additional stu
dy is needed of uncertainties and limitations of this proposed approac
h. Future applications may be found in green engineering design and ma
nufacturing changes.