Sustainability science is an emerging scientific field that aims to address the environmental problems facing contemporary societies. This article explores the relationship between the wickedness level of these problems and the research stances and methods scientists use to address them. It reviews a sample of 17 research projects addressing diversely wicked environmental problems, all of which originate in the same distributed network of research infrastructures in France. We distinguished between the political complexity and the cognitive complexity of the problems addressed and between the collaborative pluralism and the methodological pluralism of the projects. While we expected overall positive relationships between these paired aspects, we found positive but, at best, weakly significant correlations between cognitive complexity and political complexity, between methodological pluralism and collaborative pluralism, and between problem wickedness and project pluralism. We identified three research stances: a correspondence between project pluralism and problem complexity; reductionism, when methodological or collaborative pluralism was lower than expected; and integrationism, in the opposite case. We found that project pluralism tended to increase and the latitude of choice between research stances tended to decrease according to problem wickedness. Addressing highly wicked problems thus seems to significantly constrain research stances and methods. Our empirical data also suggested the possible influences of project duration and leadership on project pluralism. This article thus clarifies the factors that influence how sustainability science is concretely carried out and the constraints that addressing highly wicked problems places on scientists.