Collaboration has become a common alternative to top-down approaches for managing natural resources. Little is known, however, about how science is used in collaborative processes. We analyze how the use of science is affected by the structure of collaborative processes, with specific attention to the means by which stakeholders are engaged, the source of information gathered, and how decisions are made. We use case survey methods (a form of meta-analysis that recodes qualitative text as categorical variables) to analyze 30 case studies of collaborative management of marine and coastal ecosystems in the U.S. Preliminary findings indicate that increased stakeholder participation improves the use of science (as measured by seven dependent variables). These findings conflict with arguments that science would play a lesser role as stakeholder participation is broadened beyond the scientific community. The findings also indicate that face-to-face communication among stakeholders and increased contribution of scientific information from academic sources promotes the use of science in collaborative processes. On the less positive side, however, we find that science is more likely to be incorporate into final decisions when acknowledgements of scientific uncertainty are underplayed. Thus, while science appears to play a greater role in multi-stakeholder collaborative processes than commonly thought, the findings are mixed in terms of how well it is used from a scientific perspective. The paper will conclude with recommendations regarding how the structure of collaborative processes can be improved to enhance the appropriate use of science for environmental conservation.