The uranium-series isotope signatures of the suspended and dissolved load of rivers have emerged as an important tool for understanding the processes of erosion and chemical weathering at the scale of a watershed. These signatures are a function of both time and weathering-induced fractionation between the different nuclides. Provided appropriate models can be developed, they can be used to constrain the residence time of river sediment. This chronometer is triggered as the bedrock starts weathering and the inferred timescale encompasses the residence time in the weathering profile, storage in temporary sediment deposits (e.g. floodplain) and transport in the river. This approach has been applied to various catchments over the past five years showing that river sediments can reside in a watershed for timescales ranging from a few hundreds of years (Iceland) to several hundreds of thousands of years (lowlands of the Amazon). Various factors control how long sediment resides in the watershed: the longest residence times are observed on stable cratons unaffected by glacial cycles (or more generally, climate variability) and human disturbance. Shorter residence times are observed in active orogens (Andes) or fast-eroding, recently glaciated catchments (Iceland). In several cases, the residence time of suspended sediments also corresponds to the time since the last major climate change. The U-series isotope composition of rivers can also be used to predict the river sediment yield assuming steady-state erosion is reached. By comparing this estimate with the modern sediment yield obtained by multi-year sediment gauging, it is clear that steady-state is seldom reached. This can be explained by climate variability and/or human disturbance. Steady-state is reached in those catchments where sediment transport is rapid (Iceland) or where the region has been unaffected by climate change and/or human disturbance. U-series are thus becoming an important tool to study the dynamics of erosion.