Age at first full-term pregnancy (FTP) has long been thought to be the major reproductive risk factor in breast cancer but a Norwegian study suggested that age at last FTP might be more important. In Norway "high parity" means 4 or more deliveries. Does this finding hold in an area with a much broader distribution of parity? Data from a case-control study done in 1980-82 in Fortaleza and Recife, two cities in Brazil's impoverished north-east, have been used to explore further the influence of age at last FTP. The cases were 509 women with histologically diagnosed breast cancer who were matched with hospital controls for age and area of residence. The analysis was based on case-control pairs interviewed by the same person. High breast cancer risk was associated with low parity; after adjustment for parity, breast cancer risk was related both to late age at first FTP (odds ratio [OR] 1.21 for each 5 year increase, p = 0.008) and to late age at last FTP (OR 1.24, p = 0.0007). However, multivariate analysis revealed that the effect of age at last FTP dominated that of age at first FTP: once age at last FTP was taken into account the effect of age at first FTP was no longer significant (OR 1.08, p = 0.38) while the association with parity became more striking. These results challenge the view that age at first FTP is the principal reproductive variable related to breast cancer risk. Moreover, they suggest that high parity is protective independent of ages at first and last FTP. Given recent worldwide reductions in fertility rates, breast cancer incidence may be expected to increase. Balancing that may be the willingness of some women to complete their families by, say, age 35 if they were to be told that this might reduce their risk of breast cancer.