During the past decade, toxicological research has been dominated by t
wo themes: investigations to elucidate the mechanisms of action of tox
icants and studies to provide information to support improved assessme
nts of human health risks. The conduct of mechanistic investigations w
as given an early impetus by advances in biochemistry and cell biology
and, more recently, by related advances in molecular biology. Researc
h to provide information for improved human health risk assessments wa
s stimulated by the 1983 NAS/NRC report that provided a codified struc
ture for conducting risk assessments. At first glance, it would appear
that the two themes are closely related and, indeed, should represent
parts of a joined theme. However, examination of the toxicology/risk
assessment literature of the past decade indicates that this has not b
een the case. Reports of mechanistic studies infrequently indicate how
the information can be used to provide improved estimates of human ri
sk from exposure to toxicants. If reference is made, it is usually qua
litative in nature. Neither is examination of the risk assessment lite
rature reassuring. Mechanistic studies may be cited; however, the fina
l step of the process, risk characterization, is usually dominated by
the use of default options grounded in conservative interpretations of
generic scientific knowledge. Two examples are reviewed that stand ou
t as illustrations of how mechanistic information can be used to make
a difference in risk assessments: (1) consideration of the alpha(2u)-g
lobulin-mediated mechanism for evaluating male rat data for relevance
in assessing human risks of renal cancer and (2) the use of DNA-protei
n cross-links as an internal dose metric in cross-species extrapolatio
n of nasal cancer risks from inhaled formaldehyde. This paper reviews
past experience on these topics and suggests a strategy for increasing
the use of mechanistic information in risk assessments. A key compone
nt of the strategy is to use the risk assessment process to identify r
esearch needs/opportunities that, if addressed, will reduce the use of
default options, thereby reducing the uncertainties in risk assessmen
ts. Another component bf the strategy is to identify a few chemicals a
nticipated to exert their effect via different mechanisms and whose me
chanisms of dosimetry and disease pathogenesis can be investigated in-
depth within a risk assessment framework; this identification will cre
ate prototype approaches as alternatives to the use of default options
that have major impact on the outcome of the risk assessment process.