This paper examines apparent tensions between "science-based," "precautionary," and "participatory" approaches to decision making on risk. Partly by reference to insights currently emerging in evolutionary studies, the present paper looks for ways to reconcile some of the contradictions. First, I argue that technological evolution is a much more plural and open-ended process than is conventionally supposed. Risk politics is thus implicitly as much about social choice of technological pathways as narrow issues of safety. Second, it is shown how conventional "science-based" risk assessment techniques address only limited aspects of incomplete knowledge in complex, dynamic, evolutionary processes. Together, these understandings open the door to more sophisticated, comprehensive, rational, and robust decision-making processes. Despite their own limitations, it is found that precautionary and participatory approaches help to address these needs. A concrete framework is outlined through which the synergies can be more effectively harnessed. By this means, we can hope simultaneously to improve scientific rigor and democratic legitimacy in risk governance.