Soil/solution partitioning of trace metals (TM: Cd, Co, Cr, Cu, Ni, Sb, Pb and Zn) has been investigated in six French forest sites that have been subjected to TM atmospheric inputs. Soil profiles have been sampled and analysed for major soil properties, and CaCl2-extractable and total metal content. Metal concentrations (expressed on a molar basis) in soil (total), in CaCl2 extracts and soil solution collected monthly from fresh soil by centrifugation, were in the order: Cr > Zn > Ni > Cu > Pb > Co > Sb > Cd , Zn > Cu > Pb = Ni > Co > Cd > Cr and Zn > Ni > Cu > Pb > Co > Cr > Cd > Sb , respectively. Metal extractability and solubility were predicted by using soil properties. Soil pH was the most significant property in predicting metal partitioning, but TM behaviour differed between acid and non-acid soils. TM extractability was predicted significantly by soil pH for pH < 6, and by soil pH and Fe content for all soil conditions. Total metal concentration in soil solution was predicted well by soil pH and organic carbon content for Cd, Co, Cr, Ni and Zn, by Fe content for Cu, Cr, Ni, Pb and Sb and total soil metal content for Cu, Cr, Ni, Pb and Sb, with a better prediction for acidic conditions (pH < 6). At more alkaline pH conditions, solute concentrations of Cu, Cr, Sb and Pb were larger than predicted by the pH relationship, as a consequence of association with Fe colloids and complexing with dissolved organic carbon. Metal speciation in soil solutions determined by WHAM-VI indicated that free metal ion (FMI) concentration was significantly related to soil pH for all pH conditions. The FMI concentrations of Cu and Zn were well predicted by pH alone, Pb by pH and Fe content and Cd, Co and Ni by soil pH and organic carbon content. Differences between soluble total metal and FMI concentrations were particularly large for pH < 6. This should be taken into account for risk and critical load assessment in the case of terrestrial ecosystems.