This article describes a technique for the measurement of total and diffusible F content of serum, at clinical significant concentrations of F (1–10 μM). The proposed procedure avoids the interference of unknown serum components with the ion‐specific electrode. Sample F is concentrated fivefold through distillation of hydrofluoric acid (Taves' method). Ionic fluoride is presented to the electrode in a simple solution at concentrations within the linear response of the electrode. Average recoveries of F from serum or its ultrafiltrate were 96 ± 7% (21%) and 97 ± 12% (53%) (mean ± SEM [CV]), respectively. With four replicates of each sample, the technique produce within‐run standard deviations of 0.6 μM and 2.2 μM at 1 and 10 μM F, respectively. Total precision assessment gave standard deviations of 0.6 μM and 2.6 μM at 1 and 10 μM F, respectively. The fasting serum F levels of normal climacteric women, 45 to 65 years, showed an asymmetric distribution. The data obtained started at the detection limit of the technique (0.1 mM). The 75 percentile was 1.85 μM for total and 0.5 μM for diffusible F. In patients (n = 25) treated with NaF (30 mg F/day) the fasting levels of total serum F (4.5 ± 1.7 μM) did not differ from those of diffusible F (4.2 ± 1.5 μM). In patients (n = 50) treated with sodium monofluorophosphate (15 mg F/day) the fasting levels of total and diffusible serum F were 6.5 ± 1.7 μM and 0.5 ± 0.03 μM, respectively.In conclusion, this paper establishes the presence of two fractions of serum fluorine: diffusible and nondiffusible (or protein bound) and describes a technique for their clinical estimation. In untreated subjects and in patients receiving NaF, the former fraction contains ionic fluoride. In patients treated with MFP, diffusible serum fluorine is composed by ionic fluoride and low molecular weight, peptide‐bound, acid‐labile fluorine. J. Clin. Lab. Anal. 13:151–157, 1999. © 1999 Wiley‐Liss, Inc.