The development of in vitro protocols able to discriminate skin irritants from non-irritants integrates the toxicologists' needs for reliable and robust in vitro tools for screening test substances. Based on EpiSkin test method, validated by ESAC (ECVAM Scientific Advisory Committee) in April 2007 as the Draize skin irritation replacement reference test method, we present and discuss here the results obtained by adapting protocols to the SkinEthic Reconstructed Human Epidermis (RHE) model. The main adaptations of the validated reference protocol consists in a modulated exposure time (15, 42 or 60min) followed by a rinsing step and a 42h post-incubation period before quantitative measurement of cell viability by MTT reduction. The results obtained with a set of 39 test substances allowed to determine a prediction model with a cut-off of 50%. The best reliability was obtained with the proposed "42 bis" (42min+42h) test method. An overall accuracy of 85% was reached when testing the 20 ECVAM selected reference test substances. The performance of this optimized test method was confirmed by its higher robustness compared to other proposed protocols. As such, none of test substances showed a standard deviation above 18%. This optimized skin irritation protocol has thus been established according to the ECVAM intra-laboratory minimum performance standards.