Steady state plasma carbamazepine (CBZ), carbamazepine epoxide (CBZE), and carbamazepine diol (CBZD) concentrations were quantified by HPLC in 121 specimens obtained from two groups of epileptic patients: 78 receiving CBZ monotherapy (I), and 43 receiving CBZ and valproic acid (VPA) (II). The differences of drug/metabolite ratios and concentration/dose (mumol/L/mg/kg/day or 1/clearance) ratios were calculated as a measure for the influence of VPA on CBZ metabolism. Results as means +/- SD were CBZ/CBZE 5.85 +/- 3.91 (I) vs. 4.22 +/- 1.57 (II), p < 0.02; CBZ/CBZD 2.94 +/- 1.94 (I) vs. 2.82 +/- 1.15 (II); CBZE/CBZD 0.53 +/- 0.24 (I) vs. 0.71 +/- 0.32 (II), p < 0.001. Concentration/dose ratios: CBZ 2.32 +/- 1.58 (I) vs. 3.04 +/- 1.41 (II), p < 0.05; CBZE 0.41 +/- 0.20 (I) vs. 0.73 +/- 0.28 (II), p < 0.001; CBZD 0.82 +/- 0.35 (I) vs. 1.22 +/- 0.70 (II), p < 0.001. Drug/metabolite relationship data seem to support the concept for VPA as a selective inhibitor of epoxide hydrolase, but concentration/dose ratios indicate a reduced clearance for the parent drug, and especially for its two metabolites. This latter finding is in a controversy with the former concept. In addition, a considerable age-dependency of the influence of VPA on CBZ metabolism was found: compared to monotherapy, drug/metabolite and concentration/dose ratios were most changed in children. We assume that VPA is probably not a selective inhibitor of epoxide hydrolase, and affects nonspecifically all steps of the epoxide-diol pathway of CBZ metabolism.