This study describes a GC-MS method for the simultaneous determination of androstenedione (AD), 11beta-hydroxyandrostenedione (11beta-OHAD), and testosterone (TS) in human plasma. [19,19,19-(2)H(3)]Androstenedione (AD-(2)H(3)), 11beta-hydroxy-[1,2,4,19-(13)C(4)]androstenedione (11beta-OHAD-(13)C(4)), and [1,16,16,17-(2)H(4)]testosterone (TS-(2)H(4)) were used as internal standards. Pentafluoropropionic (PFP) derivatization with good GC behavior was employed for the GC-MS analysis of the three steroids. The detection limit of the present GC-MS-SIM method was found to be 1 pg per injection for AD (S/N ratio=4.5), 5 pg for 11beta-OHAD (S/N ratio=5.0), and 1 pg for TS (S/N ratio=4.4), respectively. Calibration curves were linear from 0.22 to 2.80 ng/mL (r=0.9998) for AD, from 0.56 to 3.19 ng/mL (r=0.9996) for 11beta-OHAD, and from 2.05 to 10.3 ng/mL (r=0.9996) for TS. The intra- and inter-day assay reproducibilities in the amounts of the three androgens determined were in good agreement with the actual amounts added, the relative errors (R.E.) were -3.1 to 2.4%. The inter-assay relative standard deviation (R.S.D.) was less than 5.3%. The present method provides a sensitive and reliable technique for the simultaneous determination of AD, 11beta-OHAD, and TS in plasma. The method can be applied to pharmacokinetic and metabolic studies of androgens with a particular interest in evaluating the conversion of AD to 11beta-OHAD and the interconversion of AD and TS in humans.