In an earlier study, we found that rice (Oryza sativa) grown in nutrient solution well-supplied with Zn preferentially took up light (64)Zn over (66)Zn, probably as a result of kinetic fractionation in membrane transport processes. Here, we measure isotope fractionation by rice in a submerged Zn-deficient soil with and without Zn fertilizer. We grew the same genotype as in the nutrient solution study plus low-Zn tolerant and intolerant lines from a recombinant inbred population. In contrast to the nutrient solution, in soil with Zn fertilizer we found little or heavy isotopic enrichment in the plants relative to plant-available Zn in the soil, and in soil without Zn fertilizer we found consistently heavy enrichment, particularly in the low-Zn tolerant line. These observations are only explicable by complexation of Zn by a complexing agent released from the roots and uptake of the complexed Zn by specific root transporters. We show with a mathematical model that, for realistic rates of secretion of the phytosiderophore deoxymugineic acid (DMA) by rice, and realistic parameters for the Zn-solubilizing effect of DMA in soil, solubilization and uptake by this mechanism is necessary and sufficient to account for the measured Zn uptake and the differences between genotypes.