The particulate contaminant flux (φ) across a river section is the product of the discharge (Q), the suspended particulate matter concentration (Cs) and the content in contaminant adsorbed on the particles (Ci.) High temporal resolution sampling of Q and Cs can provide continuous time series with a controlled uncertainty for both parameters. In contrast, Ci measurements are often infrequent, due to analysis costs and sampling difficulties, resulting in a larger uncertainty in the flux estimation. The influence of the temporal sampling of Ci was studied using a large data set on particles content for many contaminants collected 4 times per year from 2000 to 2009 at 17 sites throughout the Rhône River catchment by the regional water authority. About 300 individual molecules were analysed, however many of them were never quantified. Information about the molecule hydrophobicity and comparison with data collected in the liquid phase may allow to optimize the number of analysis. Then we studied discontinuity over time of the Ci values for some contaminants and highlighted changes in extraction protocols and reported limits of quantification, that were linked with a change of the analytical laboratory. Intercomparison exercises between laboratories and a complete documentation on analytical methods and quality control results are therefore of utmost importance. Finally, we observed that events with the highest Cs values were unsurveyed, which probably leads to an underestimation of the Ci variations. Thus extrapolation of Ci can lead to huge uncertainties on calculated fluxes for the most contributing events. Therefore, the sampling effort should be focused on the most sediment-laden events, relying on an automatic sampler, to estimate Ci variability with the hydrological cycle.