For the study of in-vitro skin penetration of candidate drugs, excised animal skin is frequently used as a replacement for human skin. Reconstructed human skin or epidermis equivalents have been proposed as alternatives. We compared the penetration properties of human, pig and rat skin with the Graftskin LSE (living skin equivalent) and the Skinethic HRE (human reconstructed epidermis) models using four topical dermatological drugs (salicylic acid, hydrocortisone, clotrimazole and terbinafine) with widely varying polarity. In agreement with published data, pig skin appeared as the most suitable model for human skin: the fluxes through the skin and concentrations in the skin were of the same order of magnitude for both tissues, with differences of at most two- or fourfold, respectively. Graftskin LSE provided an adequate barrier to salicylic acid, but was very permeable for the more hydrophobic compounds (e.g. about 900-fold higher flux and 50-fold higher skin concentrations of clotrimazole as compared to human skin), even more than rat skin. In the case of the Skinethic HRE, we found similar concentrations of salicylic acid as in human skin and an approximately sevenfold higher flux. In contrast, the permeation of hydrophobic compounds through the epidermal layer was vastly higher than through split-thickness human skin (up to a factor of about 800). To conclude, currently available reconstituted skin models cannot be regarded as generally useful for in-vitro penetration studies.