A retrospective study of the population pharmacokinetics of tiagabine was performed from sparse data collected in a multicentre clinical trial in patients with newly diagnosed partial seizures. The purpose was to estimate the inter patient variability and to study the influence of various demographic, environmental and pathophysiological parameters on the pharmacokinetics of tiagabine in patients on monotherapy. A total of 593 plasma concentrations from 130 patients dosed with 2.5, 5, 7.5 or 10 mg tiagabine twice daily were used for modelling. A one-compartment open model with first-order absorption and elimination was fitted to the concentration-time data using the NONMEM program. Selection of covariates was initially performed using stepwise linear regression analyses. The selected covariates were incorporated in the population model and the importance of each covariate was investigated by means of backwards elimination. A one-compartment model with first-order absorption and elimination adequately described the tiagabine concentration-time profile. The apparent clearance as well as the apparent volume of distribution were both significantly correlated to body height in a nonlinear relationship. No other demographic, environmental or clinical chemical parameters were identified as covariates although only a few pathological values of the latter were present in the data. The mean values of CL/f was 6.10 l/h, of V/f was 62.0 l and of k(a) was 1.25 h(-1) for a subject of 170-cm height. The population half-life was 5.72 h. The apparent clearance and volume of distribution of tiagabine in epilepsy patients on monotherapy were both dependent on body height. Prospective studies are required in order to reveal if dose adjustments based on body height will result in improved therapeutic outcome.