The stereoselective determination of stereoisomers in biological samples provides vital information on stereospecific metabolism and pharmacokinetic profiles of the drugs. Despite the unique advantage and the great success of normal-phase (NP) HPLC for the separations of drug stereoisomers using polysaccharide-type chiral stationary phases (CSPs), the technique is rarely applied to quantitative HPLC–MS–MS bioanalysis. This is, at least in part, due to the incompatibility between the usual mobile phase ( n-hexane or n-heptane) in normal-phase HPLC and the MS ionization sources which poses a potential detonation hazard. An environmentally friendly and nonflammable alternative solvent, ethoxynonafluorobutane (ENFB), was reported previously to potentially provide an ideal solution for combining the powers of stereoselective NP chromatographic separation and MS–MS detection. In this study, a stereoselective NP-HPLC–MS–MS method was developed using ENFB to quantify a pair of Bristol Myers Squibb (BMS) proprietary drug stereoisomers and their ketone metabolite for an in vitro study, which demonstrated, for the first time, the practical applicability and utility of ENFB for bioanalysis in pharmaceutical industry. The effects of different organic modifiers and temperature, as well as the comparison between ENFB and the usual solvent, heptane, for the separation, are discussed. The resolution of the stereoisomers was achieved using 63% of 3:1 mixture of ethanol and methanol with 37% ENFB on a Chiralpak AD-H column at 50 °C. High sensitivity was obtained using the MS–MS detection in the positive ion atmospheric pressure chemical ionization (APCI) mode. The lower limit of quantitation (LLOQ) for the first stereoisomer and the ketone metabolite was 5 ng/mL, and was 10 ng/mL for the second isomer in the human liver microsome–potassium phosphate buffer matrix. The linear dynamic range of 5–1000 ng/mL for both isomers and 10–1000 ng/mL for the metabolite were demonstrated with R 2 ≥ 0.997. The precision of the analysis was <5% R.S.D. at or above the nominal concentration of 80 ng/mL, and <20% R.S.D. at 8 ng/mL. The mean bias was less than 15%. Extraction recovery and acceptable matrix interference were demonstrated using one isomer and the ketone, and better than 75% recovery and less than 25% ion suppression or interference were found. The method was successfully implemented for an in vitro intrinsic metabolic clearance study.