To develop a composite index to describe the overall breast-feeding performance of infants < 6 months of age; and, using this index, to identify the factors associated with poor breast-feeding practices and the association between breast-feeding and infant morbidity.

The 2003 Demographic and Health Survey was a multi-stage cluster sample survey of 4320 households in Timor-Leste which covered 573 infants aged < 6 months. Breast-feeding Performance Index (BPI) was constructed by allocating one point for each of seven infant feeding practices: first suckling within an hour of birth; absence of prelacteals; non-use of feeding bottles; current breast-feeding; not receiving liquids; not receiving formula or other milk; and not receiving solids in the last 24 hours. BPI was treated as the dependent variable in univariate and multivariate analyses to identify the factors associated with poor breast-feeding.

Exclusive breast-feeding rate was 29.9%. The BPI (mean 4.4, standard deviation 1.77) was categorised as low, average and high according to tertiles. Multivariate analysis indicated that infants from the richest households were 1.70 (95% confidence interval (CI) 1.04–2.77) times more likely to have ‘low BPI’ than the poorest. Maternal BMI < 18.5 kg m− 2 was predictive of poor breast-feeding (odds ratio = 1.79; 95% CI 1.27–2.52). In the ‘low’ BPI group, the incidence of diarrhoea (13.4%) and acute respiratory infections (20.7%) during the previous two weeks was significantly higher than in ‘average’ (4.3 and 9.3%) and ‘high’ BPI groups (4.6 and 5.5%).

Creating a composite index to assess the overall breast-feeding performance among infants < 6 months of age is feasible. BPI can be effectively used to identify target groups for breast-feeding promotion interventions.