In a case-control study of 451 women with breast cancer and 451 population-based controls from metropolitan Adelaide, Australia, the risk of breast cancer was studied in relation to intake of dietary fiber and various fiber components. Intakes were estimated by means of a self-administered quantitative food-frequency questionnaire, and risks of breast cancer were estimated for each quintile of fiber density (intake/megaJoule of total energy intake) relative to an arbitrarily assigned risk of unity for women in the lowest quintile of fiber density. There were highly significant reductions in relative risk associated with increasing intake of a number of components of total dietary fiber. In women with the highest 20% estimated dietary densities of total non-starch polysaccharides (NSP), the relative risk of breast cancer was 0.46, and a test for trend across the quintiles was significant at p < 0.001. Separate risk estimations for various fiber components defined by the Englyst method of analysis revealed highest risk reductions associated with high densities of mannose from insoluble NSP and of glucose from soluble NSP. For many fiber components the point estimates of relative risk were quite similar both for pre- and for post-menopausal women, although results for the latter (who comprised over two thirds of all cases) were statistically more significant. The study provides strong support for recent conjecture that foods rich in dietary fiber may be protective against breast cancer.