In recent decades, U- and Th- series nuclides have been successfully used to determine weathering rates in soils and regoliths. However, the redistribution of the nuclides within secondary minerals and/or organic matter may interfere with the recorded U- and Th-series weathering signals and can result in inaccurate and/or poorly constrained weathering rate determinations. In addition, the migration of radionuclides in the environment represents a major current issue. The objective of this study, therefore, was to evaluate the meaning of bulk soil data for weathering investigations and to assess the processes and timescales of U, Th and Ra redistribution in a forested-soil profile, both at the bulk soil scale level and for separated mineral fractions. This study was carried out at the INRAE-Andra experimental beech-forested site of Montiers (Meuse, France), which was developed on acidic detrital sediments inherited from tropical paleosoils erosion. U- and Th-series activity nuclides were measured in bulk soil samples and in separated Fe nodules and clay-size (<2 mu m) fractions. Furthermore, the analyses of atmospheric particles allow discussing the potential influence of atmospheric inputs.While bulk soil disequilibria suggested negligible to moderate mobility of the nuclides, an important transfer between mineral fractions was highlighted; this resulted in contrasting redistributions of U, Th and Ra within soil fractions. Fe-oxide nodules controlled the overall mobility of U on the bulk soil scale, while the fine <2 mu m fraction controlled the mobility of Ra. A capture of U-234 by Fe-oxide nodules (mainly comprised of goethite) was observed within the entire soil profile, most likely occurring for several hundred ky. In the shallow soil layers, a leaching of U from the <2 mu m fraction that was not re-adsorbed onto Fe-oxides occurred. In contrast, Fe-oxides appeared to retain all U mobilized from the <2 mu m fraction in the deep soil layers, resulting in negligible U mobility and an apparent "closed-system" at the bulk soil scale. This transfer, however, seemed to occur over shorter timescale than the one in the shallowest soil layers. No mobility of Ra-226 was observed at the bulk soil , scale, despite the large release of this nuclide from Fe-oxides. The measured ((210)pb/Ra-226) activity ratios suggested that up to 50% of Ra-226 in Fe-oxides is available for desorption. However, the <2 mu m fraction strongly readsorbed the released Ra and prevented the loss of this nuclide at the bulk soil scale. Considering the short halflife of Ra-226, such transfer from Fe-oxides to the <2 mu m fraction occurred over a significantly shorter timescale than U mobilization. In addition, the analysis of the short-lived Ra-228 isotope emphasized the flux involved in the biological cycle and the potential impact of vegetation on the soil radionuclide redistribution over a short timescale.Our results suggest that the fine <2 mu m fractions might be suitable for probing the residence time of clay minerals in soils by the U-series dating method, if the soil profile can be assumed to consist of homogenous inherited clays.