Ecosystem science deals empirically with processes that drive biophysical attributes of large landscapes. Ecosystem management should use ecosystem science to facilitate sustained access to ecosystem goods and services (sensu Lubchenco et al. 1991) that humans need or value highly, but polemics about ecosystem management threaten its usefulness and a unified definition and protocol for the concept has proven elusive. In recent workshops and papers about ecosystem management (e.g., Christensen et al. 1996), we have observed three recurring themes: (1) concurrent evaluation of human actions on availability or quality of multiple biological, social, and ecological resources or values should occur at some relevant landscape scale, (2) multidisciplinary, sound science should form the information base from which alternative management actions are derived, and (3) uncertainty or risk analysis should guide selection of management actions to sustain ecosystem function. Taken individually, these themes are not new. Federal land managers have embraced a multiple-use concept since the 1960s, if not earlier. Reliance on multidisciplinary collaborations is a fundamental feature of contemporary ecosystem science (e.g., the LongTerm Ecological Research Program of the National Science Foundation, Franklin et al. 1990). Risk analysis is commonly used in many aspects of natural resource management (e.g., western USA wildfire management). Why, then, is the concept of adaptive ecosystem management so controversial? We believe that the basic three themes rarely are integrated and applied in a regional ecosystem context due to the lack of a unified protocol for implementation of these themes at a landscape scale. We examine herein the three themes that we think underpin the concept of ecosystem management. We then offer a general protocol that links science and public process to multiple resource management objectives in an ecosystem context. Our model is based upon: (1) our belief that shared environmental and social goals, relevant science, and an informed public are requisite for sustaining regional ecosystem processes that provide essential environmental goods and services (e.g., food, fiber, clean water, and air), (2) our belief that the management process must be iterative in order to allow the integration new knowledge and changing societal goals, and (3) our experience in providing ecosystem science as underpinning for management actions that attempt to sustain the pristine attributes of the Flathead Lake catchment, a 22241-km2 montane ecosystem in northwestern Montana, USA, and southeastern British Columbia, Canada.