A modification of the model of Pike et al. (Nature 1983;303: 767-70) was applied to 91,523 women in the Nurses' Health Study who did not report prevalent cancer initially and who were followed for 14 years (1,212,855 person-years and 2,341 incident breast cancers). The model took into account current age, age at all births, age at menopause, and age at menarche in predicting the annual and cumulative incidence of breast cancer. The authors found that ages both at first birth and at subsequent births have long-term influence on breast cancer incidence. The incidence density for parous women was greater than for nulliparous women for 20-30 years after the time of the first birth. However, cumulative incidence up to age 70 years was about 20% lower, 10% lower, or 5% higher for parous versus nulliparous women if their first birth was at age 20, 25, or 35 years, respectively. The authors also observed a significantly lower incidence after each additional birth as well as after menopause for women of the same age. Overall, the effect of reproductive factors (other than age at menarche) appears to influence cumulative incidence to age 70 years by a maximum of approximately 50% when women with multiple births with an early age at first birth are compared with women with a single birth at a late age.