A structure-activity relationship model for prediction of contact allergenic potential of chemicals had previously been developed. The model had been shown to be able to classify known allergens and nonallergens using data on physicochemical and reactivity parameters of functional groups by discriminant two-value multiple regression analysis. To investigate the model, six selected chemicals which had not been previously investigated for allergenicity were studied with both the model and a murine local lymph-node assay. The same compounds were predicted to be allergens (3-bromo-2-coumaranone, 1-nitrocyclohexene and alpha-acryloyloxy-beta, beta-dimethyl-beta-butyrolactone) and nonallergens (1-carbethoxy-4-piperidone, 6,7-dimethoxy-2-tetralone and 9-acetylanthracene) by both the model and the local lymph-node assay.