The dose-related effects of the fiber-rich isolate, soft white wheat bran (SWWB), and the pure fiber, cellulose, on N-nitrosomethylurea (NMU)-induced mammary tumorigenesis was assessed in F344 female rats. SWWB (45% total dietary fiber, TDF) was added to the AIN-76A high-fat diet at 9, 12, 15 and 18%; cellulose (98% TDF) was added to the same diet at 4.5, 6, 7.5 and 9%, to give equivalent amounts of TDF. The experimental diets were fed 3 days post-NMU and continued for a period of 25 weeks, at which time the experiment was terminated and tumors enumerated. It was found that significant inhibition of mammary carcinoma occurred only at 9% SWWB, non-significant inhibition occurred at 12% SWWB, and no inhibition was seen at higher doses. Cellulose-fed animals exhibited consistently higher tumor yields regardless of dose. The difference in tumor yields between the 9% SWWB group and the remaining seven groups was attributable to an increased incidence in tumors characterized histologically as intraductal proliferation and ductal carcinoma in situ in the latter. Analysis of blood, urine and fecal estrogens was conducted to test whether dietary fiber exerted its tumor-inhibiting effect by altering the enterohepatic recycling of estrogens. Although SWWB, in general, lowered urinary estrogen excretion, increased fecal estrogen excretion and lowered blood estrogens, there was no consistent correlation between the amount of SWWB consumed, estrogen status and tumor yields. These results suggest that (i) wheat bran fiber at 9%, or minor constituents associated with it, contain anti-promoting properties that cellulose lacks; (ii) SWWB appears to exert its effects by suppressing the clonal expansion phase of mammary carcinogenesis; (iii) there is an upper limit (12-15% w/w) to the protective effects of SWWB; and (iv) the effects of SWWB on mammary tumorigenesis may not be attributed to alterations in the enterohepatic recycling of estrogens.