Physico-chemical descriptors of drug molecules are often not adequate in predicting their oral bioavailability. In vitro methods can be useful in evaluating some of the different factors contributing to bioavailability. While physical parameters such as drug solubility may effect oral bioavailability, in most cases, the major determining factors are likely to be metabolism, and absorption at the intestinal level. Metabolism may be preabsorptive, as with peptides, or during absorption, particularly as a result of the activity of the intracellular enzyme CYP3A4. Absorption may be transcellular (membrane diffusion, carrier-mediated, endocytosis) or paracellular, while p-glycoprotein activity in the apical cell membrane may limit bioavailability by expelling drugs from the mucosal cells. Knowledge of the absorption mechanism is important in determining formulation strategies. The different in vitro techniques used to study absorption have advantages and disadvantages. Ussing chambers can be useful to measure bidirectional transport, but most studies use simple salt media, and full tissue viability is doubtful. Caco-2 cell monolayers are human cells, but the system is static, and gives very low rates of transport, and exagerated enhancement of the paracellular route compared with small intestine. The rat everted gut sac incubated in tissue culture medium maintains tissue viability and gives reliable data, although it is a closed system. In situ perfusion gives no information on events at the cellular level, and absorption may be reduced by anaesthesia and surgical manipulation. In vivo perfusion in man, with multichannel tubes, gives valuable data, but is not practical for screening. Pharmacokinetic modelling can also give useful data such as the existence of different absorption sites. Permeability values from the literature show that for small hydrophilic molecules, which pass by the paracellular route, the improved everted sac gives values close to those for humans, while values with Caco-2 cells are orders of magnitude lower.