Soil erosion and both its on-site and off-site impacts are increasingly seen as a serious environmental
problem across the world. The need for an improved evidence base on soil loss and soil redistribution
rates has directed attention to the use of fallout radionuclides, and particularly 137Cs, for documenting
soil redistribution rates. This approach possesses important advantages over more traditional means of
documenting soil erosion and soil redistribution. However, one key limitation of the approach is the
time-averaged or lumped nature of the estimated erosion rates. In nearly all cases, these will relate to the
period extending from the main period of bomb fallout to the time of sampling. Increasing concern for
the impact of global change, particularly that related to changing land use and climate change, has
frequently directed attention to the need to document changes in soil redistribution rates within this
period. Re-sampling techniques, which should be distinguished from repeat-sampling techniques, have
the potential to meet this requirement. As an example, the use of a re-sampling technique to derive
estimates of the mean annual net soil loss from a small (1.38 ha) forested catchment in southern Italy is
reported. The catchment was originally sampled in 1998 and samples were collected from points very
close to the original sampling points again in 2013. This made it possible to compare the estimate of
mean annual erosion for the period 1954e1998 with that for the period 1999e2013. The availability of
measurements of sediment yield from the catchment for parts of the overall period made it possible to
compare the results provided by the 137Cs re-sampling study with the estimates of sediment yield for the
same periods. In order to compare the estimates of soil loss and sediment yield for the two different
periods, it was necessary to establish the uncertainty associated with the individual estimates. In the
absence of a generally accepted procedure for such calculations, key factors influencing the uncertainty
of the estimates were identified and a procedure developed. The results of the study demonstrated that
there had been no significant change in mean annual soil loss in recent years and this was consistent
with the information provided by the estimates of sediment yield from the catchment for the same
periods. The study demonstrates the potential for using a re-sampling technique to document recent
changes in soil redistribution rates.