Objective: In the treatment of epilepsy, as in all areas of drug therapy, generic formulations are becoming more important as pressure to reduce drug costs increases. In clinical practice, the interchangeability of reference and generic preparations is often assumed. However, in individual patients with a narrow therapeutic range, switching to another formulation of an antiepileptic agent with low water solubility and nonlinear pharmacokinetics may lead to breakthrough seizures or toxicity. The purpose of this study was to examine the bioequivalence and clinical effects after switch of equal daily dosages of two commercially available sustained-release preparations of carbamazepine (CBZ). Patients and Study Design: This was a nonblinded intra-individual trial in 14 patients aged 18 to 52 years with focal epilepsy who had been receiving monotherapy with the reference preparation of sustained-release carbamazepine (mean dosage 1985 mg/day given twice daily) for at least 35 days. After a minimum of 3 further days, the reference preparation (denoted CBZ-R) was immediately replaced by an identical dosage of the generic preparation (denoted CBZ-G). Under steady-state conditions, for each preparation we determined area under the serum concentration-time curve (AUC), maximum serum concentration (C), minimum serum concentration (C) and peak-trough fluctuation (PTF) of carbamazepine and carbamazepine 10,11-epoxide. Results: One patient dropped out because of intolerable adverse effects with CBZ-G. For carbamazepine, the data for the remaining 13 patients, expressed as CBZ-G/CBZ-R, were: AUC 111.5% [90% confidence interval (CI) 105.6-117.8%]; PTF 90.9% (90% CI 73.4-112.8%); C 110.1% (90% CI 100.4-117.0%). These data indicate a higher bioavailability for CBZ-G. For carbamazepine 10,11-epoxide, about 20% higher average concentrations were measured with CBZ-G. Applying the usual inclusion rule [90% CI within the range 80 to 125% (AUC) or 70 to 143% (PTF and C) of the reference preparation], the two formulations can be considered as bioequivalent. In contrast, we observed marked adverse effects, such as dizziness, nausea, ataxia, diplopia and nystagmus, in eight of the remaining 13 patients after switching to CBZ-G. Conclusions: Proof of bioequivalence between reference and generic preparations of antiepileptic drugs does not mean that they are freely interchangeable. Generic formulations with proven bioequivalence to branded preparations can be used, for example, at the beginning of treatment or in poorly controlled patients with serum concentrations in the mid range. The results demonstrate that the usual rules for bioequivalence and the range of acceptability for preparations of carbamazepine are problematic. [ABSTRACT FROM AUTHOR]