The ECVAM-funded skin irritation validation study (SIVS) was initiated in 2003, with the aim to evaluate whether the EpiDerm, EPISKIN and the SIFT alternative methods were able to reliably identify skin irritant and non-irritant chemicals, and could therefore be candidates for replacing the rabbit Draize test for skin irritation. The primary goal of the study was to evaluate the predictive capacity of the assays with regard to the EU classification system, which employs the risk phrases, "R38", for skin irritants, and "no label" for non-irritants. A secondary objective was the retrospective analysis of the data, to assess whether the in vitro tests would be able to discriminate between strong irritants (category 2), mild irritants (category 3) and non-irritants (no category), as defined by the OECD and United Nations proposal for a Globally Harmonised System (GHS) for the classification and labelling of dermal irritancy. A Chemicals Selection Sub-Committee (CSSC) was appointed to identify test chemicals to be used in the SIVS, for which existing, high quality in vivo data were available, with which to correlate the in vitro measurements. Since chemicals from the European Centre for the Ecotoxicology and Toxicology of Chemicals (ECETOC) database of reference chemicals for skin irritation/skin corrosion had been extensively used in preceding studies, the CSSC made use of novel sources for potential test chemicals. The first source of chemicals screened was the New Chemicals Database (NCD), which is the central archive within the EU notification scheme for 'new' commercial chemicals. Data registered in the NCD originate from standard assays, submitted in compliance with the legislation which regulates the marketing of industrial chemicals, and are subject to quality assurance by the competent authorities of the EU Member States. In addition, to obtain 'existing' chemicals which were readily available from major manufacturing and/or distribution sources, additional databases were surveyed, such as the Toxic Substance Control Act (TSCA) database maintained by the US Environmental Protection Agency (EPA), and the ECETOC database, with the exclusion of the chemicals used in the previous optimisation and prevalidation phases. A total of approximately 3500 chemicals from the NCD and 1600 from the additional databases were screened. Pre-determined selection criteria were applied, primarily to ensure the quality of the in vivo data and the practicability of their use in testing. Overall, the number of chemicals fulfilling the CSSC selection criteria was found to be limited, particularly in the case of GHS category 2 chemicals. However, a total set of 60 chemicals were selected and proposed to the Management Team of the SIVS for independent coding and supply to the participating laboratories. The selected chemicals: i) represented statistically justified sample sizes for distinguishing R38 from no-label chemicals; ii) provided a balanced representation of the three GHS categories, to allow for the post hoc evaluation of the performance of the assays for that classification system; and iii) acknowledged, to a certain degree, the large prevalence known to exist for chemicals which have oedema and erythema scores of 0. The selected chemicals represented a variety of molecular structures, functional chemical groups, and effect and use categories, as well as a wide range of physico-chemical properties. They represented a challenging set of chemicals, relevant to current industrial commerce, with which to validate the alternative methods.