According to OECD guideline 405 revised in 1987 Draize eye tests need not be performed for severely irritating and corrosive chemicals if results from 'well-validated alternative studies' are presented. In 1988 a validation study on alternatives to the Draize eye test was started in Germany to establish 'well-validated alternative methods' for this purpose. During database development, the last stage of the validation programme, 136 chemicals from the German chemical industry were classified in a blind trial with the 3T3 cell neutral red/kenacid blue cytotoxicity assay and the hen's egg chorioallantoic membrane (HET-CAM) test using fertile chicken eggs. The major goal of this stage of validation was to demonstrate the feasibility and limitations of the two alternative methods. Chemicals were, therefore, selected as representatives of chemical structural groups as well as of physicochemical and toxicological properties. In addition, some of the chemicals were chosen because they were of interest to the cosmetic and detergent industries. Draize eye testing data in vivo were provided by industry. In contrast to data from a previous interlaboratory assessment trial, it was impossible to correlate cytotoxicity data to the EEC classification for in vivo eye irritation. However, seven of 10 severely irritating chemicals (EEC labelling R-41) could be identified correctly in the HET-CAM assay, whereas test conditions of the study described here did not allow identification of irritating chemicals (EEC labelling R-36). The HET-CAM test is, therefore, fulfilling the criteria of a 'well-validated alternative method' according to OECD guideline 405 and should be incorporated into eye irritation testing at the earliest possible stage to reduce effectively the suffering of rabbits in the Draize eye test. Although an 80% correct prediction of 'non-labelled' chemicals in the HET-CAM test is encouraging, for safety assessment of non-irritant chemicals, for use as cosmetic formulations, for example, both government and industry will accept an in vitro assay only if its prediction of the absence of irritant properties is 100% correct.