In terrestrial ecosystems, plant-available pools of magnesium and calcium are assumed to be stored in the soil as exchangeable cations adsorbed on the surface of mineral and/or organic particles. The pools of exchangeable magnesium and calcium are measured by ion-exchange soil extractions. These pools are sustained in the long term by the weathering of primary minerals in the soil and atmospheric inputs. This conceptual model is the base of input-output budgets from which soil acidification and the sustainability of soil chemical fertility is inferred. However, this model has been questioned by data from long-term forest ecosystem monitoring sites, particularly for calcium. Quantifying the contribution of atmospheric inputs, ion exchange and weathering of both primary, secondary and non-crystalline phases to tree nutrition in the short term is challenging. In this study, we developed and applied a novel isotopic dilution technique using the stable isotopes of magnesium and calcium to study the contribution of the different soil phases to soil solution chemistry in a very acidic soil. The labile pools of Mg and Ca in the soil (pools in equilibrium with the soil solution) were isotopically labeled by spraying a solution enriched in 26Mg and 44Ca on the soil. Labeled soil columns were then percolated with a dilute acid solution during a 3-month period and the isotopic dilution of the tracers was monitored in the leaching solution, in the exchangeable (2 sequential 1 mol L-1 ammonium acetate extractions) and non-crystalline (2 sequential soil digestions: oxalic acid followed by nitric acid) phases. Significant amounts of Mg and Ca isotope tracer were recovered in the non-crystalline soil phases. These phases represented from 5% to 25% and from 24% to 50%, respectively, of the Mg and Ca labile pools during the experiment. Our results show that non-crystalline phases act as both a source and a sink of calcium and magnesium in the soil, and contribute directly to soil solu