The genotoxicity and carcinogenicity data from in vitro and in vivo studies conducted during preclinical safety assessment of doxorubicin (DOXO), epirubicin (EPI) and idarubicin (IDA), are reviewed. The genotoxicity assays included a) gene mutation in Salmonella typhimurium with 5 tester strains; b) gene mutation in the V79 mammalian (lung) cell line; c) chromosome aberrations in human lymphocytes cultured in vitro; and d) chromosome aberrations in mouse bone marrow cells after intravenous (i.v.) administration in vivo. The long-term toxicity studies in the rat included a) single dose administration (3 mg/kg DOXO, 3.6 EPI and 0.75 IDA) to female rats of two different age groups, i.e. younger (7 weeks old at dosing) and older (13 weeks old), followed by one-year observation; and b) multiple dose administration to male and female rats (7 weeks old at dosing), consisting of i.v. administration of 0.25, 0.5 and 1 mg/kg DOXO or EPI and 0.06, 0.125 and 0.25 mg/kg IDA, once every 3 weeks for 10 cycles, followed by 18 months of observation. The genotoxicity studies revealed activity in gene mutation assays in bacterial and mammalian cells, and in chromosome aberration assays in human lymphocytes in vitro and in mouse bone marrow in vivo. In the two long-term studies in the rat, only mammary tumors were present. This finding was expected and, according to the literature, can be considered as species specific and not directly compound-related. The lack of tumor induction at the usual target organs for DNA reactive compounds, which are almost the same as those considered as target organs in anthracycline-exposed animals, indicates that the type and the extent of DNA damage precludes stimulation for proliferation and induction of neoplasia. Although an epigenetic mechanism can be hypothesized, support for such a mechanism is lacking.