To report the relationship between maternal prenatal intention to breastfeed and the actual initiation and duration of breastfeeding.Pregnant women resident within Avon, UK, expected to give birth between 1 April 1991 and 31 December 1992 were recruited in a longitudinal cohort study. Main outcome measures included maternal infant feeding intention (breastfeed, breast and bottle feed, bottle feed, or uncertain) at 32 wk of pregnancy: intention in the first week, intention for the rest of the first month and intention in months 2 to 4; initiation and duration of breastfeeding up to six months.Data were available on 10,548 women. Prenatal intention to breastfeed had an influence on both initiation and duration of breastfeeding. Of the women intending to bottle feed from birth, only 3.4% initiated breastfeeding compared with 96.6% of women planning to breastfeed for at least four months. At six months postpartum, the mean duration of breastfeeding for women intending to breastfeed for at least five months was 4.4 mo (95% CI 4.3, 4.4), compared with 2.5 mo (95% CI 2.4, 2.6) for women with a prenatal intention to breastfeed for only one month. Logistic regression, using intended duration as the only explanatory variable, correctly predicted 91.4% of breastfeeding initiation and 72.2% of infant feeding at six months.This large population-based study confirms the strength of the relationship between maternal prenatal intention to breastfeed and both breastfeeding initiation and duration. Maternal intention was a stronger predictor than the standard demographic factors combined. This should be taken into account in future research, and trials should be undertaken to establish whether interventions could alter maternal intention and thereby increase rates of breastfeeding initiation and duration.