With the exception of the Draize Test, the guinea-pig test methods currently accepted by regulatory authorities worldwide are well able to predict the potential of a material to cause skin sensitization. Nevertheless, (a) some methods are more sensitive than others (e.g. adjuvant tests are generally more sensitive than non-adjuvant tests); (b) methods cannot be sufficiently standardized to give full reproducibility of results between laboratories; and (c) most methods are based on subjective visual grading of skin reactions--difficulties thus arise when testing coloured or irritant materials. Laboratories must be able to show the sensitivity of the method(s) they use by demonstrating that positive reactions occur with mild/moderate contact allergens rather than the strong/extreme sensitizers currently recommended in certain guidelines, specifically in the EEC Test Method. The sensitivity of the adjuvant tests is such that it is possible to halve the minimum number of animals required by present regulatory guidelines without compromising the capacity of the tests to detect weak/mild sensitizers. A similar review has not yet been made for non-adjuvant tests. Alternative test methods, including some recently developed mouse models, offer several advantages, including more objective endpoints. These tests have not been extensively validated and this precludes their use at present for regulatory purposes other than to confirm the sensitization potential of a material. Two new test methods using mice, the Mouse Ear-swelling Test and the Local Lymph Node Assay, appear promising. They should undergo rigorous interlaboratory testing to determine their sensitivity and specificity. In vitro methods do not represent a viable alternative in the foreseeable future. An approach using quantitative structure-activity relationships is the most likely route to a non-animal model, but this will require considerable research, development and validation. Human sensitization tests have generally not been used for the classification of substances as non-sensitizers. This is because of an absence of internationally agreed test protocols, the lack of positive controls and because the methods for establishing the sensitivity of human tests are less developed than for animal tests. Nevertheless, for products for which direct human contact is intended, predictive tests in human volunteers can be considered. The EEC Directive for the Classification, Packaging and Labelling of Dangerous Substances provides a reasonable approach to the evaluation of skin sensitizers.(ABSTRACT TRUNCATED AT 400 WORDS)