Abstract The vertical variability in mineralogical, chemical and isotopic compositions observed in large river suspended sediments calls for a depth-integration of this variability to accurately determine riverine geochemical fluxes. In this paper, we present a method to determine depth-integrated chemical particulate fluxes of large rivers, based on river sampling along depth-profiles, and applied to the Amazon Basin lowland tributaries. The suspended particulate matter (SPM) concentration data from depth-profiles is modeled for a number of individual grain size fractions using the Rouse model, which allows to predict the grain size distribution of suspended sediment throughout the whole river cross-section. Then, using (1) the relationship between grain size distribution and the Al/Si ratio (2) relationships between the Al/Si ratio and the chemical concentrations, the chemical composition of river sediment is predicted throughout the river cross-section, and integrated to yield the depth-integrated chemical particulate flux for a number of chemical elements (e.g. Si, Al, Fe, Na, REEs, …). For elements such as Al, Fe, REEs, Th, the depth-integrated flux is around twice as high as the one calculated from river surface sample characteristics. For Na and Si, the depth-integrated flux is three times higher than the “surface” estimate, due to the enrichment of albite and quartz at the bottom of the river. Depth-integrated 87Sr/ 86Sr composition of suspended sediment, also predictable using this method, differs by more than 10 −3 from the surface sample composition. Finally, potential implications of depth-integrated estimates of Amazon sediment chemistry are explored. Depth-integration of particulate 87Sr/ 86Sr isotopic ratios is necessary for a reliable use of Sr isotopes as a provenance tracer. The concept of steady-state weathering of a large river basin is revisited using depth-integrated sediment composition. This analysis shows that, in the Amazon Basin river, the previously observed discrepancy between (1) weathering intensities of channel surface sediment and (2) silicate-derived dissolved fluxes is only slightly accounted for by the vertical variability of suspended sediment weathering intensities. This observation confirms that most large rivers basins are not eroding at steady-state.