The purpose of this study was to evaluate the potential value of different epithelial cell culture systems as in vitro models for studying corneal permeability. Transformed human corneal epithelial (HCE-T) cells and Statens Serum Institut rabbit corneal (SIRC) cells were cultured on permeable filters. SkinEthic human corneal epithelium (S-HCE) and Clonetics human corneal epithelium (C-HCE) were received as ready-to-use systems. Excised rabbit corneas (ERCs) and human corneas (EHCs) were mounted in Ussing chambers, and used as references. Barrier properties were assessed by measuring transepithelial electrical resistance, and by determining the apparent permeability of markers with different physico-chemical properties, namely, fluorescein, sodium salt; propranolol hydrochloride; moxaverine hydrochloride; timolol hydrogenmaleate; and rhodamine 123. SIRC cells and the S-HCE failed to develop epithelial barrier properties, and hence were unable to distinguish between the permeation markers. Barrier function and the power to differentiate compound permeabilities were evident with HCE-T cells, and were even more pronounced in the case of C-HCE, corresponding very well with data from ERCs and EHCs. A net secretion of rhodamine 123 was not observed with any of the models, suggesting that P-glycoprotein or similar efflux systems have no significant effects on corneal permeability. Currently available corneal epithelial cell culture systems show differences in epithelial barrier function. Systems lacking functional cell-cell contacts are of limited value for assessing corneal permeability, and should be critically evaluated for other purposes.