One feature of the Globally Harmonized System of Classification and Labelling of Chemicals (GHS) is that testing is not required; however, classification of mixtures based on pH and/or the individual ingredients and their respective levels in the mixture can result in over-classification in some instances. SC Johnson (SCJ) utilizes in vitro methods to classify mixtures, but there is not a single validated in vitro assay to predict potential skin irritation across all categories. SCJ investigated a number of commercially available in vitro skin irritation assays including the EpiDerm Time-to-Toxicity (ETT) as well as the validated EpiDerm Skin Irritation Test (SIT). The ETT assay utilizes the same cell-based system as the EpiDerm SIT; however, the ETT uses multiple exposure times to determine the ET50 (50% cell viability) value, whereas the SIT evaluates viability after a single exposure time. Samples with diverse chemistries and historical in vivo data were selected to be run in the ETT assay. The ET50s were analyzed to determine if thresholds could be established to permit classification of skin irritation across all GHS categories. The results of this work indicate a clear distinction exists between the ET50 values of corrosive mixtures compared with other categories; however, the data do not permit a distinction between ET50 values for irritant and non-irritant classifications. Mixtures predicted to be corrosive based on an extreme pH value underwent titratable acid/alkaline reserve (TAR) analysis and were tested in the EpiDerm Skin Corrosivity and Corrositex® assays. This work identified that Corrositex® is the most predictive in determining a UN packing group for mixtures with a compatible chemistry and expected corrosive classification based on weight of evidence. SCJ was able to develop a weight of evidence approach to address skin irritation classification in commercial mixtures as well as identify the shortcomings of some of the available in vitro skin irritation models. Additional research and validation is needed to further refine the prediction accuracy of the available assays for skin irritation classification.