This study evaluated management decision making as representatives from government agencies and conservation nongovernmental organizations, ranchers, and interdisciplinary researchers worked within the Collaborative Adaptive Rangeland Management (CAMP) experiment to 1) prioritize desired ecosystem services; 2) determine objectives; 3) set stocking rates, criteria for livestock movement among pastures, and vegetation treatments; and 4) select monitoring techniques that would inform decision making. For this paper, we analyzed meeting transcripts, interviews, and focus group data related to stakeholder group decision making. We find two key lessons from the CARM project. First, the CAM process makes visible, but does not reconcile differences between, stakeholder experiences and ways of knowing about complex rangeland systems. Second, social learning in CAM is contingent on the development of trust among stakeholder and researcher groups. We suggest future CAM efforts should 1) make direct efforts to share and acknowledge managers’ different rangeland management experiences, epistemologies, and knowledge and 2) involve long-term research commitment in time and funding to social, as well as experimental, processes that promote trust building among stakeholders and researchers over time.