While researchers and decision-makers increasingly recognize the importance of public participation in environmental decision-making, there is less agreement about how to involve the public. One of the most controversial issues is how to involve citizens in producing scientific information. Although this question is relevant to many areas of environmental policy, it has come to the fore in watershed management. Increasingly, the public is becoming involved in the sophisticated computer modeling efforts that have been developed to inform watershed management decisions. These models typically have been treated as technical inputs to the policy process. However, model-building itself involves numerous assumptions, judgments, and decisions that are relevant to the public. This paper examines the politics of public involvement in watershed modeling efforts and proposes five guidelines for good practice for such efforts. Using these guidelines, I analyze four cases in which different approaches to public involvement in the modeling process have been attempted and make recommendations for future efforts to involve communities in watershed modeling.