The uptake of manganese by maize roots was monitored in vivo by 31P nuclear magnetic resonance (NMR) spectroscopy and a quantitative analysis was developed on the basis of the line broadening of the vacuolar orthophosphate (P1) signal. The line broadening, which was followed indirectly by measuring changes in the reciprocal peak height of the P1 signalin fully relaxed spectra, was found to depend on pH and P1 concentration, as well ason the presence of organic acids, but for P1 concentrations in the millimolar range the method was sensitive to Mn2+ concentrations as low as 0·1–1 μM. A linear relation was established between the reciprocal peak height of the vacuolar P1 signal observed in vivo and the total manganese content of the tissue determined subsequently by atomicabsorption. However, the paramagnetic contribution to the line widthobserved in vivo was much smaller than expected from measurements on simple solutions and freeze-thaw extracts and it was concluded that less than 5% of the manganese taken up by the root tissue was present in the vacuoles as soluble Mn2+. The ability to detect the free pool of divalent manganese is one of several advantages of the 31P-NMR method relative to the analogous1H-NMR method based on the interaction between manganese and water; and the non-invasive nature ofthe method, coupled with the potential to distinguish the cytoplasmic and vacuolar manganese fractions, allows the NMR method to complement the information obtained by atomic absorption.