Recent advances in techniques for culture of human skin cells have led to their potential for use as in vitro models for skin irritation testing to augment or replace existing rabbit skin patch tests. Our work is directed towards the development of cultured human skin cells, together with endpoints that can be linked to in vivo mechanisms of skin irritation, as in vitro models for prediction of human skin irritation, and for study of mechanisms of contact irritant dermatitis. Three types of commercial human skin cell cultures have been evaluated, epidermal keratinocytes and partially or fully cornified keratinocyte-dermal fibroblast co-cultures. Human epidermal keratinocyte cultures (Clonetics) were treated with product ingredients and formulations, and the extent of cell damage was assessed by incorporation of the vital dye neutral red. Cell damage correlated with human skin patch data for ingredient chemicals with the exception of acids and alkalis, but did not correlate with skin irritation to surfactant-containing product formulations. Cultures of human skin equivalents were evaluated as potential models for measurement of responses to test materials that could not be measured in the keratinocyte/neutral red assay. We developed a battery of in vitro endpoints to measure responses to prototype ingredients and formulations in human epidermal keratinocyte-dermal fibroblast co-cultures grown on a nylon mesh ('Skin2' from Advanced Tissue Sciences) or on a collagen gel ('Testskin' from Organogenesis). The endpoints measure cytotoxicity (neutral red and MTT vital dye staining, lactate dehydrogenase and N-acetyl glucosaminidase release, glucose utilization) and inflammatory mediator (prostaglandin E2) release. Initial experiments indicate a promising correlation between responses of the Skin2 model to prototype surfactants and in vivo human skin irritation. The responses of Testskin cultures to acids and alkalis help to prove the concept that a topical application model can measure responses to these materials. These results suggest that human skin cell models can provide useful systems for preclinical skin irritation assessments, as alternatives to rabbits, for at least certain classes of test substances.