The accumulation of genetic changes is considered as the main factor that determines the development of cancer. Recent progresses in genetics and molecular biology led to the discovery of many new molecular markers and to the development of techniques able to monitor these markers. As a consequence, molecular epidemiology has emerged as a powerful approach to study the ternary relationship between the environment, the behaviour and the genetic predisposition of each individual. Susceptibility to cancer is determined at different levels such as the genetic polymorphism of enzymes involved in the activation and detoxification of carcinogens, the polymorphism of genes that maintains the genome stability, like those involved in DNA repair or recombination processes, and finally the polymorphism in oncogenes or tumour suppressor genes. Consequently, the full assessment of each individual's genetic predisposition is a long and difficult task. As the accumulation of mutations in somatic cells integrates all these parameters, its measurement would facilitate the evaluation of the individual predisposition status, provided that a marker common to a large spectrum of carcinogens could be found. Our current studies on the molecular mechanisms of carcinogen-induced mutagenesis has revealed that G-rich repetitive sequences are mutational hot spots for several major classes of environmental genotoxins such as aromatic and heterocyclic amines, polycyclic hydrocarbons and oxidative agents. We thus consider the possibility that these sequences form a new class of biomarkers for carcinogen exposure. In order to validate this hypothesis, we designed a sensitive PCR-based assay able to detect specific mutations induced by a common food-borne carcinogen in the colon epithelium of rats exposed for a short period to this carcinogen. This assay is sensitive enough to allow early detection of induced mutations and therefore allows to differentiate between unexposed animal and those exposed for a period as short as 1 week.