Liquid chromatography-tandem mass spectrometry (LC-MS/MS) has been increasingly used to measure steroids in human saliva. We studied the performance of a conventional LC-MS/MS for measuring dehydroepiandrosterone (DHEA), testosterone and progesterone in human saliva. These three steroids were co-extracted by liquid-liquid extraction and derivatized. Derivatives were resolved on a C18 column and quantified using an LC-MS/MS (AB Sciex API 2000) instrument. The assay's limits of quantification were 0.03 ng/mL for all three steroids. Inter-assay coefficients of variation were 16.6–18.8% (DHEA), 12.0–15.8% (testosterone), and 12.7–19.3% (progesterone). Assay linearity analysis showed R2 of 0.9926, 0.9750 and 0.9949 for DHEA, testosterone and progesterone, respectively. No carry-over between samplings was observed. An ion-enhancement effect of 11.6% for DHEA determination and ion-suppression effects of 13.9% and 20.7% for analysis of progesterone and testosterone, respectively, were determined. No interferences by 9 steroid analogs were detected. Spiked recoveries were 85.5% (DHEA), 86.5% (testosterone), and 92.6% (progesterone). Comparison with laboratory developed test (LDT)-LC-MS/MS methods by other New York State Department of Health certified laboratories revealed R2 = 0.9425 (DHEA, LC-MS/MS = 1.0267 LDT + 21.989), R2 = 0.9849 (testosterone, LC-MS/MS = 0.9447 LDT + 9.8037), and R2 = 0.9736 (progesterone, LC-MS/MS = 1.1196 LDT + 0.0985). Reference intervals for the 3 steroids in saliva for young males and females were estimated. Results of intra-individual salivary progesterone analysis indicated that caution should be exercised when using progesterone concentrations in predicting ovulation for females who are under treatment with birth control pills/devices or has body a weight of > 90 kg. © 2021 Elsevier B.V.