BACKGROUND Poor comparability between laboratories is often observed in the measurement of HbA2. A measurement procedure of higher metrological order is needed for value assignment to a reference material that shall be used as primary calibrator. METHODS A reference measurement procedure has been developed based on isotope dilution mass spectrometry (IDMS). The α- and δ- subunits are quantified by signature peptides released by tryptic digestion of a 25 μL-blood sample. Full length U-15N-labeled HbA0 and HbA2 are used as internal standards and added to the sample at concentrations closely matching the levels of the natural forms in blood. By this, an improvement in precision could be achieved with respect to previous mass-spectrometry based methods. RESULTS Recovery of HbA2 added to a blood sample was within 102.6-105.2%. Repeatability and within-laboratory imprecision was <2.0% for two blood samples containing HbA2 at a low and a high fraction. Total combined measurement uncertainty is estimated as 5.5%. Good agreement (r = 0.998) of results was obtained in a comparison of two laboratories using the described IDMS procedure. There is good correlation between commercial analytical systems and IDMS (r = 0.975-0.989). Some of the platforms provide significantly biased results, however, which potentially could be mitigated by reference to IDMS. CONCLUSIONS IDMS holds a promise to be suitable as a reference measurement procedure for standardization of HbA2-measurements in laboratory medicine.