It has been suggested that pregnancy is associated with a short-term increased risk of breast cancer followed by a life-long protection. We studied 340 incident cases of breast cancer in relation to parity, age at first full term birth, and time since last child birth during 14 years follow-up in a prospective cohort of 29,981 Norwegian women. We found no evidence for a transient increase in risk of breast cancer subsequent to pregnancy followed by a reduction in risk of long lasting duration. However, our results indicate that up to an approximate age of 45 years, the nulliparous have a lower breast cancer risk than everparous women. Among parous women, there was an increasingly protective effect on breast cancer risk at a young age (less than 50 years) with increasing number of child births, independent of age at first birth, whereas the protection associated with an early age at first full term pregnancy may increase in importance with increasing age. This finding may reconcile conflicting reports related to these two factors. This study confirms the results of previous investigations suggesting that a "cross-over" in breast cancer incidence between nulliparous and ever-parous women appears to take place some time during the fifth decade of life.