End-member modelling of bulk grain-size distributions allows the unravelling of natural and anthropogenic depositional processes in salt marshes and quantification of their respective contribution to marsh accretion. The sedimentology of two marshes is presented: (1) a sheltered back-barrier marsh; and (2) an exposed, reinstated foreland marsh. Sedimentological data are supplemented by an age model based on lead-210 decay and caesium-137, as well as geochemical data. End-member modelling of grain-size data shows that marsh growth in back-barrier settings is primarily controlled by the settling of fines from suspension during marsh inundation. In addition, nearby active dunes deliver aeolian sediment (up to 77% of the total sediment accretion), potentially enhancing the capability of salt marshes to adapt to sea-level rise. Growth of exposed marshes, by contrast, primarily results from high-energy inundation and is attributed to two sediment-transport processes. On the seaward edge of the marsh, sedimentation is dominated by coarser-grained traction load, whereas further inland, settling of fine-grained suspension load prevails. In addition, a third, coarse-grained sediment sub-population is interpreted to derive from anthropogenic landreclamation measures, that is material from drainage channels relocated onto the marsh surface. This process contributed up to 34% to the total marsh accretion and terminated synchronously with the end of land reclamation measures. Data suggest that natural sediment supply to marshes alone is sufficient to outpace contemporary sea-level rise in the study area. This underlines the resilience potential of salt marshes in times of rising sea levels. The comparison of grain-size sub-populations with observed climate variability implies that even managed marshes allow for the extraction of environmental signals if natural and anthropogenic sedimentary processes are determined and their relative contribution to bulk sediment composition is quantified. Data series based solely on bulk sediments, however, seem to be of limited use because it is difficult to exclude bias of natural signals by anthropogenic measures.