A study has been carried out investigating four different marketed oral preparations of glibenclamide, correlating the effectiveness of the drug in these preparations in lowering plasma glucose concentrations with the in vitro dissolution of the drug, measured by the British Pharmacopoeal and Desaga methods, and the in vivo bioavailability, assessed in 12 healthy human volunteers. The two dissolution methods yielded different rank orders of ease of dissolution of the drug from the various preparations; the findings of neither dissolution method correlated adequately with the results of the in vivo bioavailability studies, which correctly predicted the abilities of the preparations to reduce plasma glucose concentrations. Relative to an oral glibenclamide solution the bioavailabilities of the drug from three tablet preparations were 0.69, 0.49 and 0.24. The mean elimination half-life of the drug was 1.5 h and assuming complete bioavailability of the drug from oral solution the mean systemic clearance was 0.095 l kg-1h-1, and the mean apparent volume of distribution was 0.20 l kg-1. It is concluded that it may be unsafe to use in vitro dissolution data as a basis for assessing the bioequivalences of different glibenclamide preparations intended for oral use.