Two-year protocols are commonly used for long-term assay of chemicals for carcinogenicity. However, in some cases, if animals have been exposed to tested agent from the young age and sacrificed at the age of 2, an underestimation of the carcinogenic potential of non-genotoxic agents and tumor promoters is possible. Data available show that exposure to non-genotoxic agents or tumor promoters result in tumor development in older than in young animals. Age-related accumulation of "premalignant" cells in several tissues is here discussed. Accumulation with age of spontaneous mutations or mutations evoked by endogenous mutagens can induce genome instability and, hence, increase the sensitivity to carcinogens and/or tumor promoters. It seems reasonable to recommend the use of old animals for long-term assay of compounds with suspected non-genotoxic and tumor-promoting activity. It may be important to reevaluate standard two-year protocols for long-term assay for carcinogenicity and it is recommended to keep animals for at least 30 months, better until natural death, despite non profitable from an economical point of view it is more reliable.