A skin irritation test (SIT) utilising a common protocol for two in vitro reconstructed human epidermal (RhE) models, EPISKIN® and EpiDerm™, was developed, optimised and evaluated as a replacement for the in vivo rabbit skin irritation test in an ECVAM-sponsored validation study. In 2007, both RhE models were recognised by an independent peer-review panel and the ECVAM Scientific Advisory Committee (ESAC) as validated for use with the common SIT protocol. The EPISKIN SIT was endorsed as a full replacement of the in vivo rabbit test. Since the EpiDerm SIT proved to be less sensitive than the in vivo test and the EPISKIN SIT, the test was recognised as a validated component of a tiered testing strategy, in which positive results are accepted and negative results require further confirmation. The ESAC, in its April 2007 statement, also recommended increasing the sensitivity of the EpiDerm SIT, in order to gain the full acceptance. Analysis of the EpiDerm and EPISKIN data from the ECVAM validation study indicated that the lower sensitivity of the EpiDerm SIT might be linked to the more robust barrier properties of the EpiDerm model. This hypothesis was also in line with results published previously. To overcome the relatively low sensitivity of the EpiDerm protocol as a hindrance to full regulatory acceptance, a modification of exposure conditions was introduced into the protocol to achieve better agreement with the in vivo rabbit data. In the Modified EpiDerm SIT protocol, the test chemical exposure time was increased from 15 minutes to 60 minutes. In addition, part of the exposure was performed at 37°C. When the 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT) viability assay endpoint was used for classification, a significant increase of sensitivity was obtained (86.1%), whilst maintaining the high specificity of the method (76.3%). With the change to the EU classification system, which now uses higher cut-off for the classification of irritants, the sensitivity of the Modified EpiDerm SIT increased to above 90%. The measurement of interleukin (IL)-1α release did not further contribute to improvement of the method. The results demonstrate that the modified EpiDerm SIT protocol has the required sensitivity and specificity to be accepted as a stand alone method for complete replacement of the in vivo rabbit test.