The primary goal of paleoflood hydrology is to estimate the frequency and magnitude of past floods. Botanical evidence, and particularly scars on trees, has been used repeatedly as paleostage indicators to reconstruct peak discharges and flood height. Yet, these reconstructions depend on the presence of visible scars on tree stems which tend to be masked as trees grow older. Here, we estimated flood magnitude using an alternative approach based on growth disturbances in tree-ring series, tree positions and the minimal discharge necessary to submerge the root collar of a tree as estimated by hydraulic modeling. We tested the reliability of this newly developed approach by using the traditional scar-based reconstruction as a benchmark. To this end, we sampled 60 trees showing evidence of flood damage on their stems along a 787-m long segment of the Asco river (Corsica, France). Based on 440 growth disturbances dated in tree-ring series, we reconstructed 28 floods between 1759 and 2020 and 18 during the 20th century. Using the two-dimensional Iber hydraulic model and detailed topographic data of the study site obtained from UAV imagery, we estimated that peak discharges of the 28 reconstructed events ranged between 10 and 210 m3s−1, with 200 m3s−1 being considered as the threshold for extreme floods. Not only do the scar-based and root collar submersion approaches yield similar results, findings are also clearly in line with the sparse information available from historical archives and short gauge station records on past floods. The unprecedented length and depth of the record presented here opens new avenues for climate change and flood impact research.