Animal use and welfare in toxicity testing are a major concern. These issues have been recognised by the scientific community and a number of in vitro assays, intended to reduce the use of animals, are under various stages of development However there are currently no in vitro models that can provide data directly applicable to human skin irritation. This thesis aimed to assess the suitability of the human keratinocyte as an in vitro model for predicting skin irritation potential in man. The incorporation of human tissue into a cell culture model is a major advantage of an in vitro approach as it enables direct interspecies comparison. The measurement of intracellular acid phosphatase (AP) activity has been shown to give an early indication of toxicity in rat tongue epithelial (RTE) cells. Neutral Red (NR) uptake and AP were chosen as endpoints of cytotoxicity and, in preliminary studies with SDS, were measured in cultures of human keratinocytes (HK) and 3T3 cells. AP in HK cultures exhibited a peak activity. The peak produced in 3T3 cells was negligible in comparison. AP may therefore be a specific indicator of cytotoxicity in keratinocytes. Further studies were carried out which compared in vitro AP and NR uptake data with in vivo data. Both assays exhibited good: correlations with animal rank order data but showed poor correlations with human data. The good rank correlation between the two assays and in vivo animal data implies a relationship between cytotoxicity measured in vitro, and skin irritation in animals. Cell culture techniques normally require the test material to be diluted in culture medium which precludes the testing of insoluble compounds. The NR release assay was investigated using A431 cells, as a possible model for testing insoluble cosmetic formulations. However, results compared poorly with the available in vivo human skin irritation data for the same formulations. An alternative to conventional submerged cell culture methods is the growth of keratinocytes at the air-liquid interface. Attempts were made to grow human keratinocytes at the air-liquid interface for 14 days, but results obtained were not reproducible. A cytotoxicity assay, if used in isolation from other assays, may only provide information on cell viability and therefore would provide little information on the mechanisms of irritation. Primary irritation in human and animal skin is characterised by an inflammatory reaction mediated by arachidonate metabolites and cytokines. Preliminary studies measured the release of inflammatory mediators following exposure to SDS. A cytotoxic response was confirmed by the NR uptake assay. A dose-related release of inflammatory mediator was observed although large inter-experimental variations in release were seen. Further studies were carried out using 4 more chemicals. In contrast, inter-experimental variation was slight and statistically significant increases in mediator levels were demonstrated at non-cytotoxic concentrations. The assays were also able to distinguish between non-irritants and severe irritants. It may be possible to use this information to predict, to some extent, mechanisms of inflammation. Multiple mechanisms are involved in the process of skin irritation and therefore it is unlikely that one in vitro assay alone would be able to predict human skin irritancy effectively. Data presented in this thesis imply that the combination of a cytotoxicity assay and a mediator release assay may prove useful in a battery for routine screening of compounds. The development of suitable in vitro assays, to replace animal testing, is ultimately limited by the problem of obtaining relevant in vivo data before appropriate comparisons can be made. However, inflammatory mediator release assays are based on the underlying mechanisms of inflammation and as such comparisons with in vivo data are not essential.