In clinical practice, steroid measurements are performed mainly by direct, non-extraction immunoassays adapted to high throughput, automated immunoassay platforms and employing secondary calibrators. The accuracy of such steroid immunoassays is limited by cross-reactivity with structurally related steroids and nonspecific matrix interference as well as the metrological traceability of manufacturer supplied calibrators. The accuracy of steroid immunoassay calibrators has been little investigated by independent chemical methods. Steroid concentrations of 41 calibrators (4-6 replicates per calibrator) supplied by four manufacturers for use in testosterone (T), estradiol (E2), and progesterone (P4) commercial immunoassays were measured by ultra-pressure liquid chromatography-mass spectrometry (UPLC-MS). Among 14 non-zero T calibrators, six (43 %) deviated significantly from the label concentration with 29 % outside 20 % of it. Among 14 E2 calibrators, eight (57 %) deviated significantly, whereas seven (50 %) were outside 20 % of the label concentration. Among 11 P4 calibrators, eight (73 %) deviated significantly whereas four (36 %) were outside within 20 % of the label concentration. We conclude that inaccurate calibration of manufacturer's supplied standards may contribute to inaccuracy of commercial direct steroid immunoassays.