The IPCC and other global environmental assessment processes stress the need for national scientific participation to ensure decision makers' trust in the associated scientific conclusions and political agendas. The underpinning assumption is that the relationship between scientists and decision makers at the national level is characterized by trust and interpretive synergy. Drawing on ethnographic research in Brazil, this article challenges that assumption through a case study of the policy uptake of divergent scientific interpretations as to whether or not the Amazon is a net carbon sink. It shows that the carbon sink issue became a site for struggles between important Brazilian scientists and decision-makers with central authority over the definition of the country's official position in international climate negotiations. In a geopolitically charged scientific controversy involving scientific evidence bearing on the Kyoto Protocol, Brazilian decision makers studied revealed critical distance from national scientists advancing evidence that the Amazon is a net carbon sink. As such, the decision-makers' interpretations were at odds also with dominant framings in the Brazilian media and closer to those of American scientists involved in carbon cycle research in the Amazon. Seeking to explain this disconnect, the paper discusses the divergent policy preferences of key scientists and decision-makers involved, and the correlations of these preferences with interpretations of the available scientific evidence. It identifies the continued impact of a national political tradition of limited participation in decision making and suggests that this tradition-while increasingly challenged by countervailing democratizing trends-is reinforced by key Brazilian decision makers' constructions of science as a medium through which rich countries maintain political advantage. Reflecting this, key Brazilian decision-makers justified rejecting national scientists' interpretations of the Amazon as a significant overall carbon sink by suggesting that the scientists' scientific training and associated foreign interactions bias them in favor of foreign interests, compromising their ability to accurately identify national interests. The paper situates its analysis in terms of theories of the science-policy interface and argues for greater attention to the role of culturally and politically laden understandings of science and the role of science in policy and geopolitics.