The aim of the study was to select a dissolution test method for carbamazepine (CBZ) immediate release tablets, giving the best in vitro/in vivo correlations (IVIVC) and to determine the potential of this method as an estimate for bioequivalence testing. Four 200 mg CBZ products which are sold on the Dutch market, covering the innovator and three generic products, were selected. They had been tested in a randomised, fourway cross-over bioavailability study in healthy volunteers. Their dissolution rate behaviour in vitro was investigated in two dissolution media: (1) 1% sodium lauryl sulphate in water (SLS), in accordance with the United States Pharmacopeia (USP); (2) 0.1 mol/l Hydrochloric acid in water (HC). In the bioavailability study these products had shown no large differences in the extent of absorption (AUC(0-infinity);) but large differences in absorption rate. The products now also showed large differences in dissolution rate in vitro in both dissolution media, the rank order being the same as for the absorption rate. It was concluded that the absorption rate in vivo depends on the dissolution rate in vivo. 'Level C' IVIVC according to the USP were optimised by plotting percentages dissolved on selected time points (D values) or their reciprocals (1/D values), against several pharmacokinetic parameters primarily related to the absorption phase and against AUC(0-infinity). In this way for each IVIVC the optimum D or 1/D value, was calculated. For both media no meaningful IVIVC were obtained with AUC(0-infinity), but favourable IVIVC were obtained with the parameters primarily related to the absorption phase. In the bioavailability study indicated above it was found that, among the pharmacokinetic characteristics primarily related to the absorption phase, C(max) is the most promising in expressing rate of absorption in bioequivalence testing in single dose studies with CBZ immediate release tablets. Consequently, C(max) was selected for expressing rate of absorption. The most favourable IVIVC were obtained with D(20) in SLS versus C(max). From this IVIVC and the requirements for bioequivalence (AUC(0-infinity): 0.8-1.25 and C(max) : 0.75-1.35; 90% confidence interval), a specification for dissolution testing in SLS was calculated as follows: 'after 20 minutes, 34-99% dissolved'. Owing to the fact that the rate of absorption in vivo depends on i.a. the dissolution rate in vivo, it can be concluded that with this specification bioequivalence with respect to both rate of absorption and extent of absorption is ensured. As this specification is comparable with the USP specification: 'not less than 75% dissolved after 1 h', it is concluded that the USP specification is suitable to ensure bioequivalence of CBZ immediate release tablets.