The science and management of grizzly bears (Ursus arctos horribilis) in the Rocky Mountains of North America have spawned considerable conflict and controversy. Much of this can be attributed to divergent public values, but the narrow perceptions and incomplete and fragmented problem definitions of those involved have exacerbated an inherently difficult situation. We present a conceptual model that extends the traditional description of the grizzly bear conservation system to include facets of the human domain such as the behavior of managers, elected officials, and the public. The model focuses on human-caused mortality, the key determinant of grizzly bear population growth in this region and the interactions and feedback loops among humans that have a major potential influence on bear mortality. We also briefly evaluate existing information and technical methods relevant to understanding this complex human-biophysical system. We observe not only that the extant knowledge is insufficient for prediction (and in some cases for description), but also that traditional positivistic science alone is not adequate for dealing with the problems of grizzly bear conservation. We recommend changes in science and management that could improve learning and responsiveness among the involved individuals and organizations, clarify some existing uncertainty, and thereby increase the effectiveness of grizzly bear conservation and management. Although adaptive management is a promising approach, we point out some key—as yet unfulfilled—contingencies for implementation of a method such as this one that relies upon social processes and structures that promote open learning and flexibility in all facets of the policy process.