This paper discusses EPA's acquisition and use of science in two decisions regarding National Ambient Air Quality Standards: the 1987 Revision of the NAAQS for Particulate Matter and the 1993 Decision Not to Revise the NAAQS for Ozone. In the first case, more than ten years before EPA proposed to revise the NAAQS for particulates, narrowly-scoped results of academic experiments suggesting that the agency should focus its regulatory efforts on smaller diameter suspended particulates penetrated deep into the agency, far removed from decision-makers in Washington. The particulates review was elaborate and protracted and was promoted and inhibited by multiple factors. Due to the lengthy review period, however, researchers involved in complex epidemiological studies were able to produce information which bore directly on regulator's questions prior to the final decision. Such is a rarity given the normal mismatch between the pace of regulatory decision-making and the time required to produce, analyze, and verify original scientific data. These studies observed an increase in respiratory ailments in children at particulate concentrations experienced in U.S. urban areas and suggested the lack of a discernible threshold in the relationship between particulate levels and mortality. Furthermore, as the decision was being finalized, the agency leadership was warned that forthcoming studies would probably suggest health concerns at even lower levels than the lowest end of the proposed range. In the case of ozone, political factors and a divided Clean Air Science Advisory Committee (CASAC) led to a 1992 proposal by EPA Administrator Reilly not to revise the NAAQS for ozone. Administrator Browner essentially inherited this decision during the transition period between the Bush and Clinton administrations, and although the role of science in the final decision-making was not substantive, the scientific review process was considered. The ozone case study illustrates that an elaborate and lengthy NAAQS review process is required to make science available for consideration by EPA decision-makers and that policy disagreements within CASAC provide the Administrator with a justification to not revise the NAAQS on the basis of "scientific uncertainty." Both cases provide examples of non-agency scientists operating in multiple, overlapping roles inside and outside the regulatory decision-making process. The NAAQS case studies also underscore that the Clean Air Act is based on the false scientific premise that a threshold level exists below which health effects from ubiquitous air pollutants will not be observed. As a consequence of this mistaken legislative presumption, new scientific developments inevitably point toward ever more stringent ambient standards and preordain--in principle--the outcome of periodic reviews of the scientific basis of air quality regulation. In practice, EPA's response has been to delay the inevitable.