A candidate reference measurement procedure (RMP) for measurement of unconjugated estriol in human serum has been developed and validated. The proposed method is highly reliable and uses isotope dilution coupled with liquid chromatography-tandem mass spectrometry (ID-LC-MS/MS) and requires no derivatization. An appropriate amount of serum was accurately weighed and spiked with an isotopically labeled internal standard. Unconjugated estriol and its internal standard were extracted from serum matrix using liquid-liquid extraction prior to reversed-phase LC-MS/MS. Calibrator bracketing was used to give higher specificity and accuracy for assigning serum level. The accuracy of the candidate RMP was validated by split-sample comparison to established RMPs. The lowest limit of detection (LLoD) and lowest limit of quantification (LLoQ) for developed RMP was estimated to be 0.14 nmol/L and 0.35 nmol/L, respectively. Both intra- and inter-assay imprecisions were <= 2.19% at 1.39, 17.34 and 69.35 nmol/L, respectively. Recoveries were 98.54% to 100.34% and linear response ranged from 0.35 to 173.38 nmol/L. No interference was observed. Biases were 5.6% and 2.8% against the targets of RELA2015A (3.87 nmol/L) and RELA2015B (40.62 nmol/L), respectively. Moreover, the candidate RMP was successfully applied to measure level of unconjugated estriol in serum samples of pregnant women (n = 3) and compared with two immunoassays in clinical laboratory. Our developed method is simple, accurate, and can be used as a candidate RMP to determine total unconjugated estriol level in human serum. Further improvement of certain immunoassays in accuracy and precision is needed.