BACKGROUND: Etomidate is used as a fast-acting hypnotic with few cardiovascular effects to induce anesthesia in patients with a poor cardiovascular reserve. The bispectral index (BIS) has been suggested to be a measure of the depth of anesthesia and correlates well with the level of consciousness. This study examined the population pharmacokinetics and pharmacodynamics of etomidate using nonlinear mixed effect (NONMEM) modeling and sigmoid Emax modeling. METHODS: Eighteen middle aged adults, with ASA physical status I or II, who were scheduled for elective surgery, were included. 0.2% etomidate was administerd at 150 ml/h until the patients lost consciousness. The patient recovered spontaneously until they regained consciousness, as determined by a verbal response. The BIS was determined and arterial blood samples were collected. The plasma concentrations were measured with high performance liquid chromatograhy (HPLC). NONMEM was used for population pharmacokinetic and sigmoid Emax model for pharmacodynamic analysis. RESULTS: The induction dose for the loss of eyelid reflexes was 0.38 mg/kg. The induction time from drug infusion to the loss of eyelash reflexes was approximately 3.5 minutes. This study took approximately 8.5 minutes from the start of drug infusion to the recovery of consciousness. The pharmacokinetic parameters were t_1/2α = 1.1 min, t_1/2β = 1.9 min, t_1/2γ = 106.5 min, k₂₁ = 0.36 L/min, k₃₁ = 0.009 L/min, V₁ = 6.43 L, V_area = 426 L, Cl = 2.77 L/min. The pharmacodynamics were k_eo = 0.40 L/min, CE₅₀ = 1.0 microgram/mL, E₀ = 94, E_max = 94 and γ = 1.2. The performance error for the etomidate concentration was 0.14±0.99 (typical prediction) and -0.03±0.40 (individual prediction) and -0.09±1.00 and -0.001±0.13 for the BIS score. CONCLUSIONS: When compared with other previously published data, our pharmacokinetic parameters demonstrated a shorter half lives, a larger volume of distribution, and an increased clearance with significant interindividual differences. The pharmacodynamics showed a large interindividual variability. The reason for discrepancy might be the relatively short sampling time. However, further study will be warranted to improve the model performance in the future.