What accounts for the variation in the influence of scientists in the policy-making process? Why is it that scientists sometimes appear to exercise significant autonomy in shaping policy agendas, while at other times very little? Scientists are most influential, this paper contends, when they can leverage their recognized expertise by strategically co-opting institutionalized channels of advice. This is most likely to occur in issue areas of high complexity and ambiguity when key policy makers are dependent upon scientists for their counsel. Policy entrepreneurs within competing scientific communities, prevented from accessing key decision makers, wait until windows of opportunity open to undermine the credibility of the incumbent experts, gain access to political leaders, and refocus the policy agenda. This theory is developed and tested through a case-study analysis of the nuclear test-ban debate during the Eisenhower administration from 1954 to 1958. The findings of this paper underscore the need to treat foreign policy decision making as a series of strategic interactions between multiple actors with a broader capacity to influence the policy-making process than traditionally conceived. By doing so, scholars can better understand variations in government decision making across time and issue area, providing important insights into the role of experts in a wide range of public policy domains.