This article argues that risk assessment, supposedly the scientific co
mponent of risk regulation (as opposed to risk management, the policy
component), cannot be very scientific because too little is known. Wit
hout firm scientific knowledge, risk assessment must rely on conventio
ns promulgated by bureaucrats or an the professional judgments of scie
ntific experts; such conventions and judgments reflect not only scient
ific knowledge but also policy judgments and cultural values. The inad
equacy of scientific knowledge, coupled with the lack of public trust
in government and in experts, suggests that risk regulators should be
concerned not only with creating institutional arrangements likely to
foster trust but also with creating mechanisms for providing concerned
individuals with credible reassurance. The article concludes by discu
ssing divergences between public perceptions and expert perceptions of
risks, and the weights that a democratic society should give to each
in assessing and managing risks.