We performed an analytical evaluation of a commercially available instrument for determining ionized magnesium through use of a neutral carrier, liquid-membrane-based ion-selective electrode. Reproducibility (CV 2-4%), linearity (0.30-2.50 mmol/L), lower limit of detection (0.30 mmol/L), and absence of interference from Ca2+ indicate adequate performance for measuring ionized magnesium in plasma or serum samples in the normal to high-concentration range. Sodium in excess of 150 mmol/L caused a negative bias, which can be explained by ionic strength-induced changes in activity coefficients. The use of heparin as an anticoagulant must be restricted to concentrations15 units/mL because of the binding of magnesium to heparin. The mean +/- SD concentration of ionized magnesium and its fraction of total magnesium in 76 healthy volunteers were 0.56 +/- 0.05 mmol/L and 0.65 +/- 0.04, respectively.