We investigated several formulations of the free ion activity model (FIAM) as a means of describing plant uptake of soil Cd and Zn from contaminated soils. Lolium perenne was grown on a range of urban and metal-spiked agricultural soils selected to provide a wide range of Cd and Zn concentrations, pH values and other physico-chemical properties. Plants were grown under controlled conditions and above-ground biomass was harvested at regular intervals. Concentrations of Cd and Zn in the grass were compared with estimates of metal capacity (total or radio-labile metal content in the soil) and intensity (metal concentration in the soil solution or free divalent ion activity). The results suggested that capacity terms alone were poor predictors of plant metal uptake (r2 values between 0.001 and 0.43), while metal ion intensity provided quite reasonable predictions of the variation observed for several harvests of the grass (r2=0.60–0.87). Soil solution-to-plant transfer factors were highly pH-dependent which may suggest significant competition between trace metals and protons for sorption sites on roots. However, resolution of this question was confounded because of the strong co-variance between pH and p(M2+) in the soil pore water. Thus the influence of pH could not be separated from the effect of changing metal ion activity on uptake rate. Other possible effects on metal uptake such as dilution from increased biomass during growth and competition for uptake between different metal ions (Zn vs. Cd), or with Ca2+, appeared to play very minor roles in determining bioavailability. Several formulations of the FIAM failed to provide a consistently superior prediction of metal uptake when compared to purely empirical regression with pH and p(M2+) within the range of the data used to parameterise the models.