This study aimed to develop a population pharmacokinetic-pharmacodynamic (PKPD) model for propofol using data prospectively collected from a heterogeneous group of adult, elderly, and obese patients using the bispectral index (BIS) as a pharmacodynamic guide. Adult, obese (body mass index ≥35 kg/m2 ), and elderly patients (aged >65 years) were included. Propofol infusion was started at 2000 mg/h until loss of consciousness and then guided by target BIS values (40-60). Measurements of propofol plasma concentration were performed using gas chromatography. A PKPD model was developed using NONMEM. The data set contained 423 propofol concentrations and 483 897 BIS values from 60 patients (20-92 years, 42-136 kg). A 3-compartment model was used to describe the plasma concentrations of propofol. An allometric model using lean body weight calculated by the Janmahasatian formula was found to describe the data better than total body weight for all clearance parameters, V1 and V2. An age effect was found on Q2, Q3, V1, and V3. With respect to the PD model, the use of a 2-compartment model significantly improved the model fit. Age and total body weight were included as covariates in the final pharmacodynamic model. We propose a PKPD model for propofol anesthesia with acceptable performance accuracy in a heterogeneous group of adult, elderly, and obese patients. A new method to predict propofol induction dose is presented. This method and the possibility to directly change target BIS values in opposition to the assumed target effect-site concentration constitutes certain advantages to the clinical practice. © 2019, The American College of Clinical Pharmacology.