A process-based knowledge of the sorption mechanisms in soils is a prerequisite for the prediction of radionuclide transport in soils, plant availability, and leaching risk into groundwater. The present study combined batch sorption experiments of Cs+ and Sr2+ in agricultural soils of differing soil texture with numerical experiments using PHREEQC to identify key processes of sorption at different temperatures. Sorption was simulated for both radionuclides using cation exchange models. In addition, surface complexation was integrated into the reaction network for Sr2+. Our geochemical simulations identified cation exchange on illite as the dominant sorption process for Cs+. Selection of the site types in Cs+ sorption was mainly driven by Cs+ concentration. At low Cs+ concentrations the highly selective frayed edge sites dominated sorption behavior, while less affine site types on illite constituted the major sorption partners at high concentrations. We identified undefined cation exchange as the dominant sorption process for Sr2+, followed by surface complexation on organic matter. These process-based analyses were the basis for field-scale simulations to predict the leaching risk of Cs and Sr radionuclides in agricultural soils under humid climate conditions. For both soils and radionuclides, the distribution coefficients (Kd) varied distinctly with time in shallow layers due to changes in temperature, saturation, and the prevailing dominant sorption processes. During a 3-yr-simulation period, 137Cs+ migrated to depths of 3.6 cm (silty loam) and 7.6 cm (sandy loam), while 90Sr2+ migrated to depths of 15.6 cm (silty loam) and 23.6 cm (sandy loam) due to competitive sorption of infiltrating Ca2+ and Mg2+ ions reducing 90Sr2+ sorption and displacing 90Sr2+ from its exchange sites.