As products of both natural and social systems, rivers are highly complex historical objects. We show in this paper that historical analysis works on two different levels: one level, which we call "structural", shows the materiality of the riverine environment as the spatial-temporal product of natural factors and human impacts (bed and course alterations, pollution, etc.). On a second level –"semiotic" – we show that river systems are also social constructs and the subjects of ancient and diverse management practices. The quality of a river will be a function of the dialectical interaction between both levels. Historical analysis can uncover the inherited constraints that bear upon current management practices. To help substantiate this analytical framework, we analyse the case of the Moselle river in eastern France by using archival sources and statistical data. Severely impaired by industrial discharges from iron, coal and salt industries between the 1875s and the early 1980s, the waters of the Moselle became the subject of a social consensus between stakeholders that prevented the implementation of efficient pollution management policies until the 1990s. The example urges caution on the pervasiveness of participatory approaches to river management: social consensus does not necessarily benefit the environment.