This paper discusses EPA's acquisition and use of science in a decision under the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA): the 1983-84 suspensions of ethylene dibromide (EDB); and in a decision under the Toxic Substances Control Act (TSCA): the 1989 Asbestos Ban and Phaseout Rule. By requiring EPA to balance the risks and benefits of the commercial use of toxic substances, both statutes place considerable analytical burdens on the agency, though TSCA places a more substantial burden on EPA for acquiring science and demonstrating unreasonable risks. In the case of EDB, data produced outside EPA over which the agency had no control incited a public alarm. Because a senior EPA official had contaminated the agency's reservoir of public trust by cooking the scientific data to provide regulatory relief, EPA had no credibility to portray the health risks of EDB in an objective manner. In the case of asbestos, the reviewing court, despite its limited scientific capability and lack of political accountability, substituted its own science policy judgment for that of politically accountable decision-makers of the more expert administrative agency. The court was arguably invited to do so, however, by the substantial evidentiary judicial review standard specified for TSCA by the legislature. Both cases illustrate the need for and difficulty of generating and considering scientific information regarding tradeoffs among risks in environmental regulatory decision-making.