We review the scientific efforts over the last decades to reconstruct erosion from continuous alpine lake sediment records. We focused both on methodological issues, showing the growing importance of non-destructive high resolution approaches (XRF core-scanner) as well as progresses in the understanding of processes leading to the creation of an “erosion signal” in lakes. We distinguish “continuous records” from “event-records”. Both provide complementary information but need to be studied with different approaches. Continuous regionally-relevant records proved to be particularly pertinent to document regional erosion patterns throughout the Holocene, in particular applying the source to sink approach. Event-based approaches demonstrated and took advantage of the strong non-linearity of sediment transport in high altitude catchment areas. This led to flood frequency and intensity reconstructions, highlighting the influence of climate change upon flood dynamics in the mountain.The combination of different record types, both in terms of location (high vs. low elevation), sedimentology (high vs. low terrigenous contribution) and significance (local vs. regional) is one of the main outputs of this paper. It allows the establishment of comprehensive histories of NW French Alps erosion, but also and consequently, soil dynamics and hydrological patterns throughout the Holocene. We also discuss the influence of glacier dynamics, one of the major agents of erosion in the Alps.A major feature is the growing human influence upon erosion at a local scale since at least the middle of the Bronze Age (3500 cal. BP). However and according to the regional record from Lake Bourget, only few periods of rising erosion at local scales generated a regional record that can be discriminated from wetter climatic periods. Among them, the period between 200 BCE and 400 AD appeared to be marked by a generalised rise in human-triggered erosion at local scales in the northern French Alps.This review highlights the importance of modern high-resolution and interdisciplinary studies of lake sediments, in order to better understand the complex relationships between humans, climate and the Earth system in general. We strongly argue that regional integration of data is now required to move a step further. Such an integration is easier with cost- and time-effective methods as well as after a better definition of approaches and their limits. This should lead to a stronger collaboration between paleo-data producers and modellers in the near future.