Evidence from experimental animal tumor models suggests that in many instances, the identity and mechanism of activation of cellular oncogenes is a function of both carcinogen and tissue specificity. In addition, the activation of no single oncogene has yet been found to be either sufficient or necessary for tumorigenesis in any particular experimental system. A hypothesis to account for these and other molecular and biological observations of experimental tumorigenesis has been developed. The hypothesis is based on the premise that multiple tissue specific groups or pathways of oncogenes exist in each cell, and that activation of all the oncogenes in any of these alternative pathways leads to transformation. It is assumed that each oncogene (which may be a member of one or more pathways) has a spontaneous and a carcinogen specific probability of activation. The latter value will vary from carcinogen to carcinogen. By modelling the spontaneous and carcinogen specific probabilities of activation of each gene, the number and identity of genes in each pathway, and the number of pathways in a particular cell type, it is possible to calculate the relative potency of carcinogens, the percentage of tumors containing each activated oncogene, the dose-response relationship, and other parameters. Use of this hypothetical model gives results consistent with experimental observations on oncogene activation in carcinogen-induced animal tumors.