Allergic contact dermatitis (ACD) results in inflammation of the skin due to sensitization of the immunologic system to a particular substance. The sensitization process is limited by the compound's ability to both permeate and react with proteins in the integumentary system. Currently, only in vivo animal tests such as the local lymph node assay (LLNA) are recognized by regulatory authorities for risk assessment of ACD. A quantitative structure-activity relationship has been developed to predict relative potency, which allows for the prediction of relative sensitization potentials. The experimental values used in this study include EC3 values (the concentration at which the stimulation index equals 3) from LLNA tests. The predictions in this model enable categorization of the compounds into three groups on the basis of risk of sensitization and enable screening of candidate molecules using rapid SAM1 semiempirical calculations prior to animal testing. The model may also be used to reduce the number of animals subjected to testing by providing estimated concentrations required for useful data of risk assessment. The effect of averaging available literature values on predictive ability is also investigated. The model includes halogenated compounds, aromatic compounds, alcohols, aldehydes, and ketones. The computational investigation resulted in a two-descriptor model that is consistent with the assumed mechanism for sensitization.