1. The objective of this investigation was to determine quantitatively whether experimental epilepsy is associated with a change in the pharmacodynamics of benzodiazepines in vivo. For that purpose the pharmacodynamics of midazolam were quantified by an integrated pharmacokinetic-pharmacodynamic approach in three different models of experimental epilepsy: amygdala kindling, cortical stimulation and genetic absence epilepsy. 2. The time course of the EEG effect was determined in conjunction with the decline of drug concentrations after intravenous administration of 10 mg kg(-1) midazolam. The pharmacokinetics of midazolam were most adequately described by a bi-exponential equation. No influence of epilepsy on the pharmacokinetics of midazolam was observed. 3. The increase in beta activity (11.5-30 Hz) of the EEG as derived by Fast Fourier Transformation analysis was used as pharmacodynamic endpoint. For each individual rat the increase in beta activity was directly related to the concentration in blood on the basis of the sigmoidal Emax pharmacodynamic model. In all three models a significant reduction in the maximal effect was observed, in amygdala kindling 28%, in the cortical stimulation model 49% and in genetic absence epilepsy 37%. No differences in the other pharmacodynamic parameters, E0 EC50,u and Hill factor, were observed. 4. It is inferred that in three different models of epilepsy there is a similar change in GABAergic functioning which is associated with a significant reduction in the intrinsic activity of midazolam in vivo. These models provide therefore a useful basis for further studies on the mechanism of epilepsy-induced changes in pharmacodynamics of anti-epileptic drugs.