The association between soya foods and breast cancer risk was investigated in a prospective study of 34759 women in Hiroshima and Nagasaki, Japan. Women completed dietary questionnaires in 1969-1970 and/or in 1979-1981 and were followed for incident breast cancer until 1993. The analysis involved 427 cases of primary breast cancer in 488989 person-years of observation. The risk for breast cancer was not significantly associated with consumption of soya foods: for tofu, relative risks adjusted for attained age, calendar period, city, age at time of bombings and radiation dose to the breast were 0.99 (95% CI 0.80-1.24) for consumption two to four times per week and 1.07 (0.78-1.47) for consumption five or more times per week, relative to consumption once a week or less; for miso soup, relative risks were 1.03 (0.81-1.31) for consumption two to four times per week and 0.87 (0.68-1.12) for consumption five or more times per week, relative to consumption once a week or less. These results were not materially altered by further adjustments for reproductive variables and were similar in women diagnosed before age 50 and at ages 50 and above. Among 17 other foods and drinks examined only dried fish (decrease in relative risk with increasing consumption) and pickled vegetables (higher relative risk with higher consumption) were significantly related to breast cancer risk; these associations were not prior hypotheses and, because of the large number of comparisons made, they may be due to chance.