BACKGROUND We developed and evaluated a candidate reference measurement procedure (RMP) to standardize testosterone measurements, provide highly accurate and precise value assignments for the CDC Hormone Standardization Program, and ensure accurate and comparable results across testing systems and laboratories. METHODS After 2 liquid/liquid extractions of serum with a combination of ethyl acetate and hexane, we quantified testosterone by isotope-dilution liquid chromatography–tandem mass spectrometry with electrospray ionization in the positive ion mode monitoring 289→97 m/z (testosterone) and 292→112 m/z (3C13 testosterone). We used calibrator bracketing and gravimetric measurements to give higher specificity and accuracy to serum value assignments. The candidate RMP was evaluated for accuracy by use of NIST-certified reference material SRM971 and validated by split-sample comparison to established RMPs. We evaluated intraassay and interassay imprecision, measurement uncertainty, potential interferences, and matrix effects. RESULTS A weighted Deming regression comparison of the candidate RMP to established RMPs showed agreement with no statistical difference (slope 0.99, 95% CI 0.98–1.00, intercept 0.54, 95% CI −1.24 to 2.32) and a bias of ≤0.3% for NIST SRM971. The candidate RMP gave maximum intraassay, interassay, and total percent CVs of 1.5%, 1.4%, and 1.7% across the concentrations of testosterone typically found in healthy men and women. We tested structural analogs of testosterone and 125 serum samples and found no interferences with the measurement. CONCLUSIONS This RMP for testosterone can serve as a higher-order standard for measurement traceability and can be used to provide an accuracy base to which routine methods can be compared in the CDC Hormone Standardization Program.