The chemical weathering rate of silicates is a key parameter of several geochemical processes; however, the long-term field measurement of these rates remains challenging. Currently, there are several methods for quantifying field weathering rates, but each presents it's own unique set of difficulties. In this study uranium-series nuclides were used to investigate long-term biotite weathering in a soil profile. For this purpose, pseudomorph grains of biotites (including weathering products) were hand-picked from four horizons of a soil profile and from underlying granitic saprolite from the experimental BreuilChenue site (France). Unexpected behaviors of U and Th nuclides were recorded from these samples. An unambiguous leaching of Th-232 occurs during the weathering of biotite, while reduced U release relative to Th was observed, implying an efficient redistribution of U between the primary mineral and weathering products within the pseudomorphs. The measured U-series activity ratios reflect a regular and self-consistent pattern evolution. However, the ``U-series-derived-weathered-stage" is not always coherent with the weathering features of major elements and/or with the location of the samples within the soil profile. For instance, biotite from the saprolite displays U-series activity ratios typical of significantly weathered samples, while major elements present rather limited signs of weathering. These results suggest an incongruent leaching of U and Th isotopes and suggest that U- and Th-series nuclides probe some water-mineral interactions that occur before macroscopic mineral tansformation. Using an open-system leaching model, coupled (U-234/U-238), (U-234/Th-230), and (Ra-226/Th-230) disequilibria measured from the samples allow us to calculate a weathering duration range of 6-52 ky for most weathered samples of this soil profile. The biotite weathering rate deduced from these data ranges from 7.8 x 10(-17) mol.m(-2).s(-1) to 6.5 x 10(-16) mol.m(-2).s(-1), which is consistent with the range of field rates previously reported. This rate was compared to the weathering rate of biotites induced by the substitution of vegetation occuring 35 years ago in a nearby soil profile, which was about 2-3 orders of magnitude higher than the long-term rate reported above. This feature is perfectly consistent with the decrease in the weathering rate with time that was previously widely documented but that still remains debated. Our results allude to a valuable use of U-series isotopes for mineral-weathering field rate determination and report a decrease in silicate weathering rates over long(10,000s of years) and short-term (10s of years) weathering events recorded from the same site. The results show that short term laboratory experiments can be representative of field processes, and suggest that some forms of mineral weathering reactivation, similar to that observed during laboratory experiments, may naturally occur in soils in response to high frequency perturbation processes such as land cover change.