Maternal employment rates have increased rapidly in recent years and little is known about how this influences whether women start breast feeding. We examined data from the Millennium Cohort Study to determine whether a mother's employment status (full-time, part-time, self-employed, on leave, not employed/student) and employment characteristics are related to breast-feeding initiation. This analysis comprised 14 830 white mothers from Britain and Ireland (6917 employed) with singleton babies, born from 2000 to 2002. Information was obtained on infant feeding history and mother's employment when the cohort child was 9 months old. We found that women employed full-time were less likely to initiate breast feeding than mothers who were not employed/students, after adjustment for confounding factors [adjusted rate ratio (aRR) = 0.92; 95% confidence interval (CI) 0.89, 0.96]; however, there were no differences in breast-feeding initiation between mothers employed part-time, self-employed, or on leave and mothers who were not employed/students. Among employed mothers, those who returned to work within 4 months postpartum were less likely to start breast feeding than women who returned at 5 or 6 months [aRR = 0.95; 95% CI 0.92, 0.99], and women who returned within the first 6 weeks were much less likely to start breast feeding [aRR = 0.85; 95% CI 0.77, 0.94]. Mothers returning for financial reasons were also less likely to initiate breast feeding [aRR = 0.96; 95% CI 0.93, 0.99] than those who returned for other reasons. Policies to increase breast feeding should address how both the time and circumstances of a mother's return to employment postpartum influence whether she decides to start breast feeding.