To describe the duration of breast-feeding between 1990 and 2013 and to estimate the association between breast-feeding duration and sociodemographic, health and pro-breast-feeding policies and programmes in Latin American countries.

This is a cross-sectional study with data from Demographic and Health Surveys programme conducted in Bolivia, Brazil, Colombia, Peru and the Dominican Republic between 1990 and 2013. The median duration of breast-feeding was estimated by survival analysis. Information on pro-breast-feeding policies and programmes was extracted from the World on Breastfeeding Trends Initiative (WBTi) tool. The association between the duration of breast-feeding and WBTi tool score was analysed by multilevel survival regression.

Nationally representative cross-sectional survey from Bolivia, Brazil, Colombia, Peru and Dominican Republic.

We included children under 24 months of age, totalling 17 318 children.

Breast-feeding duration showed a significant increase in all countries, except the Dominican Republic. Mothers with higher schooling level (HR = 1·66; 95 % CI 1·35, 2·04), higher income (HR = 1·58; 95 % CI 1·40, 1·77) and overweight (HR = 1·14; 95 % CI 1·05, 1·23) breastfed for a shorter time. Breast-feeding in the first hour of life (HR = 0·79; 95 % CI 0·74, 0·83) was associated with increase in the duration of breast-feeding. Regarding WBTi, Peru presented the lowest score and the Dominican Republic presented the highest score. WBTi score was inversely related to the duration of breast-feeding for this set of countries (HR = 1·07; 95 % CI 1·02, 1·12).

Mothers with better socio-economic conditions and overweight breastfed for a shorter time. Breast-feeding in the first hour was associated with longer duration of breast-feeding. In this set of countries, higher scores from WBTi tool did not result in longer duration of breast-feeding.

To identify individual and contextual factors associated with the practice of exclusive breast-feeding (EBF).

We analysed 34 435 children under 6 months of age living in 111 municipalities in the state of São Paulo, south-eastern Brazil, who participated in a survey investigating feeding practices during the first year of life, carried out during the 1999 national vaccination campaign. The questionnaire employed included questions on the consumption, in the last 24 h, of breast milk, water, tea, other types of milk and other foods, in addition to mother and child characteristics. Information on the pro-breast-feeding measures implemented in the municipalities was also collected. The effects of individual and contextual characteristics on EBF were analysed using multilevel models.

The final model showed a greater chance of EBF in women with tertiary education (odds ratio (OR) = 1.91; 95% confidence interval (CI) 1.75–2.06); women aged between 25 and 29 years (OR = 1.52; 95% CI 1.41–1.63); multiparae (OR = 1.42; 95% CI 1.33–1.49); female babies (OR = 1.12; 95% CI 1.05–1.18); birth weight ≥3000 g (OR = 1.73; 95% CI 1.49–1.97); child follow-up in the private health-care network (OR = 1.10; 95% CI 1.02–1.18); and municipalities with four or five pro-breast-feeding measures (OR = 2.4; 95% CI 2.19–2.88). An analysis of the interactions between individual and contextual variables showed that the presence of at least four pro-breast-feeding measures in the municipality attenuated the risk of early termination of EBF associated with low maternal schooling and low birth weight, and transformed child follow-up in the public network into a protective factor against the early termination of breast-feeding.

The presence of measures aimed at protecting, promoting and supporting breast-feeding in the municipality had a positive influence on EBF and attenuated the impact of risk factors for the termination of breast-feeding.