The hypothesis that low body iron stores are protective against cancer whereas high body stores promote tumor occurrence was examined in the 1-methyl-1-nitrosourea (MNU)-induced experimental model for breast cancer. Twenty-one-day-old female Sprague-Dawley rats were randomized into one of three experimental groups and fed a formulation of AIN-76A diet modified to be low in iron (2 p.p.m.), or the same diet supplemented with an adequate (120 p.p.m.) or excess (1200 p.p.m.) amount of iron provided as FeSO4.7H2O. Rats were maintained on their respective diets throughout the experiment which was terminated 32 weeks post carcinogen administration. Rats were injected i.p. with either 25 mg MNU/kg body wt or the saline-solvent in which MNU was dissolved at 50 days of age. In the first 14 weeks, dietary iron deficiency resulted in a low hematocrit and a decrease in weight gain. The appearance of mammary tumors was markedly suppressed in this group compared to those given an adequate or excess level of iron. It has been reported in the literature that reduction in weight gain due to food restriction at a period immediately after carcinogen administration severely inhibits the subsequent development of tumors. Thus the low tumor incidence in the iron-deficient rats during this time frame could be attributed to the combined effects of low hematocrit and depressed weight gain. For the period between week 14 and week 32, the hematocrit in the iron-deficient animals was maintained at a normal level, and the body wt of these rats was comparable to that of the controls given an adequate level of iron. The rate of tumor appearance in the iron-deficient group during the second half of the experiment was similar to that of the iron-adequate group in the first half of the experiment. In other words, it appeared that once hematocrit and body wt gain were restored to normal in the iron-deficient animals, tumor incidence was only minimally affected by low dietary iron. In the second half of the experiment, the tumor incidence in the adequate iron group seemed to have plateaued, whereas it continued to rise in the excess iron group. Thus excess iron appears to be more prominent than iron deficiency in modification of mammary carcinogenesis, especially when the confounding effects of low hematocrit and reduced weight gain are taken into consideration in the latter case.