To examine the relationship of maternal employment characteristics, day care arrangements and the type of maternity leave pay to breast-feeding for at least 4 months. Cohort study. Babies aged 9 months in the Millennium Cohort Study, born between September 2000 and January 2002. A total of 6917 British/Irish white employed mothers with singleton babies. Mothers employed part-time or self-employed were more likely to breast-feed for at least 4 months than those employed full-time (adjusted rate ratio (aRR) and 95% confidence interval (CI) 1.30 (1.17-1.44) and 1.74 (1.46-2.07), respectively). The longer a mother delayed her return to work postpartum, the more likely she was to breast-feed for at least 4 months (P for trend < 0.001). Mothers were less likely to breast-feed for at least 4 months if they returned to work for financial reasons (aRR 0.86, 95% CI 0.80-0.93) or used informal day care arrangements rather than care by themselves or their partner (aRR 0.81, 95% CI 0.71-0.91). Mothers were more likely to breastfeed for at least 4 months if their employer offered family-friendly (aRR 1.14, 95% CI 1.02-1.27) or flexible work arrangements (aRR 1.24, 95% CI 1.00-1.55), or they received Statutory Maternity Pay (SMP) plus additional pay during their maternity leave rather than SMP alone (aRR 1.13, 95% CI 1.02-1.26). These findings were independent of confounding factors, such as socio-economic status and maternal education. Current policies may encourage mothers to enter or return to employment postpartum, but this may result in widening inequalities in breast-feeding and persistence of low rates. Policies should aim to increase financial support and incentives for employers to offer supportive work arrangements.