Basin-wide erosion rates can be determined through the analysis of in situ-produced cosmogenic nu-clides. In transient landscapes, and particularly in mountain catchments, erosion and transport processes are often highly variable and consequently the calculated erosion rates can be biased. This can be due to sediment pulses and poor mixing of sediment in the stream channels. The mixing of alluvial sediment is one of the principle conditions that need to be verified in order to have reliable results. In this paper we perform a field-based test of the extent of sediment mixing for a w42 km 2 catchment in the Alps using concentrations of river-born 10 Be. We use this technique to assess the mechanisms and the spatio-temporal scales for the mixing of sediment derived from hillslopes and tributary channels. The results show that sediment provenance and transport, and mixing processes have a substantial impact on the 10 Be concentrations downstream of the confluence between streams and tributary channels. We also illustrate that the extent of mixing significantly depends on: the sizes of the catchments involved, the magnitude of the sediment delivery processes, the downstream distance of a sample site after a con-fluence, and the time since the event occurred. In particular, continuous soil creep and shallow land-sliding supply high 10 Be concentration material from the hillslope, congruently increasing the 10 Be concentrations in the alluvial sediment. Contrariwise, a high frequency of mass-wasting processes or the occurrence of sporadic but large-magnitude events results in the supply of low-concentration sediment that lowers the cosmogenic nuclide concentration in the channels. The predominance of mass-wasting processes in a catchment can cause a strong bias in detrital cosmogenic nuclide concentrations, and therefore calculated erosion rates may be significantly over-or underestimated. Accordingly, it is important to sample as close as possible to the return-period of large-size sediment input events. This will lead to an erosion rate representative of the "mass-wasting signal" in case of generally high-frequency events, or the "background signal" when the event is sporadic. Our results suggest that a careful consideration of the extent of mixing of alluvial sediment is of primary importance for the correct estimation of 10 Be-based erosion rates in mountain catchments, and likewise, that erosion rates have to be interpreted cautiously when the mixing conditions are unknown or mixing has not been achieved. Ó 2013 Elsevier B.V. All rights reserved.