AbstractBackground: To assess the analytical validity of free testosterone (FTe) measurements, a reference measurement procedure (RMP) is required. For steroids, isotope dilution–mass spectrometry is accepted as state-of-the-art technology. Because FTe is defined as the hormone fraction in serum water in equilibrium with the protein-bound fraction, the RMP should include a physical separation step. The use of equilibrium dialysis (ED) or ultrafiltration (UF) is advocated. Our objective was to develop such a candidate RMP.Methods: We selected UF combined with isotope dilution–gas chromatography–mass spectrometry (ID-GC/MS) for direct measurement of Te in the ultrafiltrate. After optimization of the UF process, the complete procedure was validated by use of split-sample comparisons with indirect ED (iED) and symmetric dialysis (SyD).Results: The candidate RMP gave maximum within-day, between-day, and total CVs of 3.0%, 3.1%, and 4.3%. The Deming regression equations for the respective method comparisons were: UF-ID-GC/MS = 0.98(iED) − 53 pmol/L (r = 0.94; Sy|x= 42 pmol/L) and UF-ID-GC/MS = 0.92(SyD) + 21 pmol/L (r = 0.97; Sy|x= 31 pmol/L).Conclusions: We achieved the objective of a state-of-the-art candidate RMP, which agreed well with iED and SyD. However, we also demonstrated that a degree of discordance remains, which may require a decision from an authoritative organization on the recommended procedure to measure free hormone concentrations.