No simple universal test exists to predict accurately the mobility and availability of trace metals in soil. The principals behind existing tests are briefly discussed. One of the reasons for their limited success is that soil tests do not take into account the temporal and spatial variations in soil composition and hence metal speciation. In this study a simple variation on a classical technique is proposed. The. parameter measured is the distribution coefficient, Kd, defined as the ratio of concentrations in the adsorbed and solution phases. It has been obtained using a radiotracer. Variations of soil composition have been chemically simulated by addition of simple chemicals, taking particular account of changing composition in the rhizosphere. The approach has been applied to the distribution of nickel. The effect of concentration in Ni, other similar divalent cations (Co2+, Cu2+, Zn2+, Pd2+, Sn2+, Pb2+) and an indifferent cation, Ca2+, have been investigated. The adsorption isotherm was strongly nonlinear and could be fitted well to a Freundlich isotherm. Competitive adsorption led to a decrease in Ni sorption when other cations were present. Their effect depended upon their chemical similarity to Ni, in particular their softness. The adjustment of pH within a range often induced by biological activity was found to produce various effects, depending on the soil. In general, Kd increased as pH was increased towards neutrality and in some cases a marked decrease was observed at alkaline pH. The effect of addition of organic compounds was also complex and soil-dependent and could not be simply attributed to the formation of solution phase complexes. The sensitivity of Kd to the chosen parameters is discussed with respect to the likely consequences on mobility. Some likely mechanisms are briefly presented. Finally, the limitations inherent to this albeit useful approach are discussed