Conventional methods for the determination of past soil erosion provide only average rates of erosion of the sediment's source areas and are unable to determine the rate of at-a-site soil loss. In this study, we report in-situ produced cosmogenic 10Be, and 14C measurements from erratic boulders and two depth-profiles from Younger Dryas moraines in Scotland, and assess the extent to which these data allow the quantification of the amount and timing of site-specific Holocene soil erosion at these sites. The study focuses on two sites located on end moraines of the Loch Lomond Readvance (LLR): Wester Cameron and Inchie Farm, both near Glasgow. The site near Wester Cameron does not show any visible signs of soil disturbance and was selected in order to test (i) whether a cosmogenic nuclide depth profile in a sediment body of Holocene age can be reconstructed, and (ii) whether in situ10Be and 14C yield concordant results. Field evidence suggests that the site at Inchie Farm has undergone soil erosion and this site was selected to explore whether the technique can be applied to determine the broad timing of soil loss. The results of the cosmogenic 10Be and 14C analyses at Wester Cameron confirm that the cosmogenic nuclide depth-profile to be expected from a sediment body of Holocene age can be reconstructed. Moreover, the agreement between the total cosmogenic 10Be inventories in the erratics and the Wester Cameron soil/till samples indicate that there has been no erosion at the sample site since the deposition of the till/moraine. Further, the Wester Cameron depth profiles show minimal signs of homogenisation, as a result of bioturbation, and minimal cosmogenic nuclide inheritance from previous exposure periods. The results of the cosmogenic 10Be and 14C analyses at Inchie Farm show a clear departure from the zero-erosion cosmogenic nuclide depth profiles, suggesting that the soil/till at this site has undergone erosion since its stabilisation. The LLR moraine at the Inchie Farm site is characterised by the presence of a sharp break in slope, suggesting that the missing soil material was removed instantaneously by an erosion event rather than slowly by continuous erosion. The results of numerical simulations carried out to constrain the magnitude and timing of this erosion event suggest that the event was relatively recent and relatively shallow, resulting in the removal of circa 20–50 cm of soil at a maximum of ∼2000 years BP. Our analyses also show that the predicted magnitude and timing of the Inchie Farm erosion event are highly sensitive to the assumptions that are made about the background rate of continuous soil erosion at the site, the stabilisation age of the till, and the density of the sedimentary deposit. All three parameters can be independently determined a priori and so do not impede future applications to other localities. The results of the sensitivity analyses further show that the predicted erosion event magnitude and timing is very sensitive to the 14C production rate used and to assumptions about the contribution of muons to the total production rate of this nuclide. Thus, advances in this regard need to be made for the method presented in this study to be applicable with confidence to scenarios similar to the one presented here.