In the present study, a model of reconstructed human epidermis (RHE) was used as an in vitro model to discriminate the effects of Tween 80, Triton X100 and benzalkonium chloride (BC) as irritants and 1-chloro-2,4-dinitrobenzene (DNCB) as sensitizing agent. The extent of epidermal irritation and sensitization was evaluated morphologically and on the basis of intracellular and extracellular levels of interleukin-1alpha (IL-1alpha) and interleukin-8 (IL-8). The corresponding constitutive mRNA levels were quantified and the cytotoxic response was assessed by MTT assay. The RHE resembled normal human epidermis with all typical layers. The keratin 10 (K10) was typically confined in the suprabasal layers of the tissue, suggesting normal epidermal terminal differentiation. Topical application of the three surfactants resulted in significant changes of tissue morphology and was coupled with different dose-dependent decreases in cell viability corresponding to their in vivo irritant potency. The IL-1alpha release increased concomitantly with cell viability decrease, but surprisingly, the surfactants did not elicit elevated IL-8 levels. In contrast, DNCB did not induce elevated IL-1alpha release although it induced a rapid dose-dependent decrease in cell viability. On the contrary, it increased the IL-8 release. RT-PCR demonstrated the presence of mRNA for IL-1alpha as well as for IL-8. IL-1alpha was the most abundant cytokine transcript. BC, Triton X100 and DNCB upregulated IL-8 mRNA expression while only BC induced a significant increase in IL-1alpha mRNA expression. Our results showed that the production of IL-1alpha and its release into the extracellular medium was ascribed not only to direct cytotoxicity but also to the extent of tissue stimulation. Conversely, the production of IL-8 did not directly correlate with cytotoxicity but seems to be linked to the type of product applied either irritant or sensitizer. Our data demonstrate that divergent IL-1alpha and IL-8 release profiles and corresponding mRNA upregulation characterize the response to the tested irritants and DNCB. These results currently need confirmation by the introduction of more numerous known irritants and sensitizers in the tests used in the present study. However, they already suggest that it may be possible in a single assay to classify and to discriminate between irritant and sensitizing agents as a function of induced cytokine production patterns and of cell viability measurements. Copyright (C) 1999 Elsevier Science Ltd.