Soils contain a diverse and complex set of chemicals and minerals. Being an 'open system', both in the chemical and nuclear sense, soils have defied quantitative nuclear dating. However, based on the published studies of the cosmogenic atmospheric 10Be in soils, its relatively long half-life (1.5 Ma), and the fact that 10Be gets quickly incorporated in most soil minerals, this radionuclide appears to be potentially the most useful for soil dating. We therefore studied the natural variations in the specific activities of 10Be with respect to the isotope 9Be in mineral phases in eight profiles of diverse soils from temperate to tropical climatic regimes and evaluated the implications of the data for determining the time of formation of soil minerals, following an earlier suggestion [Lal et al., 1991. Development of cosmogenic nuclear methods for the study of soil erosion and formation rates. Current Sci. 61, 636-639.]. We find that the 10Be/9Be ratios in both bulk soils and in the authigenic mineral phases are confined within a narrower range than in 10Be concentrations. Also, the highest 10Be/9Be ratios in authigenic minerals are observed at the soil-rock interface as predicted by the model. We present model 10Be/9Be ages of the B-horizon and the corresponding soil formation rates for several soil profiles. The present study demonstrates that the 10Be/9Be ratios in the authigenic phases, e.g. clay and Fe-hydroxides, can indeed be used for obtaining useful model ages for soils younger than 10-15 Ma. However, the present work has to be pushed considerably further, to take into account more realistic age models in which, for instance, downward transport of 10Be and clays, and in-situ dissolution of clay minerals at depths, altering the 10Be/9Be ratios of the acidic solutions, are included. We show that in the case of younger soils (< 1 Ma) studied here, their 10Be inventories and 10Be/9Be ratios have been significantly disturbed possibly by mixing with transported soils. ?? 1997 Elsevier Science B.V.