The U.S. Department of Energy's Columbia River Comprehensive Impact Assessment (CRCIA) was an ambitious attempt to direct its cleanup of the Hanford Nuclear Reservation toward the most significant risks to the Columbia River resulting from past plutonium production. DOE's approach was uncommonly open, including tribal, regulatory agency, and other Hanford interest group representatives on the board that was to develop the assessment approach. The CRCIA process had attributes of the “analytic-deliberative” process for risk assessment recommended by the National Research Council. Nevertheless, differences between the DOE and other participants over what was meant by the term “comprehensive” in the group's charge, coupled with differing perceptions of the likely effectiveness of remediation efforts in reducing risks, were never resolved. The CRCIA effort became increasingly fragmented and the role its products were to play in influencing future clean-up decisions increasingly ambiguous. A procedural evaluation of the CRCIA process, based on Thomas Webler's procedural normative model of public participation, reveals numerous instances in which theoretical-normative discourse disconnects occurred. These had negative implications for both the basic procedural dimensions of Webler's model—fairness and competence. Tribal and other interest group representatives lacked the technical resources necessary to make or challenge what philosopher Jurgens Habermas terms cognitive validity claims, while DOE and its contractors did not challenge normative claims made by tribal representatives. The results are cautionary for implementation of the analytic-deliberative process. They highlight the importance of bringing rigor to the evaluation of the quality of the deliberation component of risk characterization via the analytic-deliberative process, as well as to the analytic component.