The possibility of the study of a quantitative correlation between short-term tests and carcinogenicity, instead of a qualitative one, is discussed. Four tests related to the target organ, rat liver, were considered: alkaline DNA fragmentation, DNA repair, DNA adducts and the formation of preneoplastic nodules. All the four tests showed a similar level of correlation with carcinogenic potency (r approximately equal to 0.4). With this level of correlation, the dispersion of the data appeared too large to offer a meaningful degree of quantitative predictivity of carcinogenicity, in reference to a single test. It appeared however, that the use of a battery of two or three independent short-term tests, with the above level of simple correlation, could generate a multiple correlation high enough to be potentially useful for some degree of quantitative predictivity of carcinogenic potency.