The relative bioavailability of four different carbamazepine products, showing large differences in in vitro dissolution profiles, was studied in healthy volunteers to correlate the occurrence of side effects with a measure of the rate of absorption in vivo for bioequivalence testing. Two of the three generic products investigated showed bioequivalence with respect to the extent of absorption with Tegretol. In vivo, the differences found in absorption rate were reflected in the occurrence of side effects, especially dizziness. As a measure for the rate of absorption, the partial AUC did not seem to be a good characteristic to test bioequivalence, as the variability is very high and dependent on the AUC taken. The Cmax/AUCpart seems more promising, especially the partial AUC directly after completion of the absorption process. The variability is low in the case of carbamazepine after a single dose. However, as long as no consensus on the use of other metrics and the objective (clinical or quality control aspects) of bioequivalence testing is reached, and no other pharmacokinetic characteristic is validated, Cmax should be the characteristic of choice for the rate of absorption in single-dose studies with carbamazepine products.