The ligand 1,4,7-triazacyclononane-1,4,7-tris(methylene methylphosphinic acid), NOTMP, was used to measure free MgII levels in blood plasma by 31P MRS. Separate resonances were observed for the free ligand and the MgII complex and the ratio of their resonance areas was used to evaluate the free, ionized MgII concentration, [Mg]free. The CaII and the ZnII complexes gave rise to separate resonances in the 31P spectrum in an aqueous sample. In human blood plasma samples, however, these resonances were never observed thus excluding the interference of these metal ions. Heparin, up to 150 units/ml, had no influence on the Mg-NOTMP equilibrium. The 31P MRS methodology was applied to twenty human blood plasma samples. Total MgII ([Mg]total), as measured by atomic absorption spectroscopy, averaged 0.85 +/- 0.12 mM while free ionized MgII ([Mg]free) measured by 31P MRS was 0.66 +/- 0.09 mM. The 31P MRS method gave inherently larger values for free ionized MgII than that reported by ion-selective electrodes (ISE). This was traced to a redistribution of existing plasma MgII species after the addition of about 2 mM of NOTMP. Calculations using existing thermodynamic data show that the ionized MgII concentration (iMg) and the concentration of MgII weakly complexed to small anions (Mg(comp)) both drop after the addition of NOTMP, with Mg(comp) dropping to negligible levels. Thus, the 31P MRS method appears to be less sensitive to variations in the concentration of weakly binding anions (bicarbonate, carbonate, chloride, lactate, phosphate, etc.) than the ISE method. Our data indicates that the difference between Mg(total), as measured by atomic absorption spectroscopy, and Mg(free), as measured by 31P MRS, provides an direct estimate of the protein bound MgII fraction.