The relationships between plant and soil systems were investigated using multivariate statistical methods and relative ionic impulsions. Soil samples were taken from three locations and wild plant species consisted of: Poa, Pteridium aquilinum, Diplotaxis, Plantago lanceolata and Trifolium repens. The content of Mg, Ca, Na, K, Mn, Fe, Pb, Co, Ni, Cu, Zn and Cr in soils and plants was determined. A five-step chemical fractionation procedure (speciation) was applied to soil samples. Total metal contents were determined and amounts extracted with HCl, EDTA and DTPA were measured. Pb, Co, Ni, Cu, Fe, Zn, Cd and Cr in soils (considered as microconstituents for plants) show a greater number of statistically significant relationships with plant contents than those shown by macroconstituents (Mg, Ca, Na, K and Mn). On the other hand, only Zn and Pb extracted with EDTA and DTPA seem to be related to metal contents in wild plants. When using relative ionic impulsions, any soil fraction (obtained from the fractionation procedure, including the soluble fraction, which is the sum of all the fractions except the residual) is suitable for the study of soil-plant system, suggesting a global balance among all the fractions. Our results suggested an active assimilation for K and a passive uptake for Na, Ca and Mn Furthermore, Fe seemed to favour the active assimilation of the other microelements.