Saliva and urine are the two main body fluids sampled when breast milk intake is measured with the 2H oxide dose-to-mother technique. However, these two body fluids may generate different estimates of breast milk intake due to differences in isotope enrichment. Therefore, we aimed to assess how the estimated amount of breast milk intake differs when based on saliva and urine samples and to explore whether the total energy expenditure of the mothers is related to breast milk output. We used a convenience sample of thirteen pairs of mothers and babies aged 2–4 months, who were exclusively breastfed and apparently healthy. To assess breast milk intake, we administered doubly labelled water to the mothers and collected saliva samples from them, while simultaneously collecting both saliva and urine from their babies over a 14-d period. Isotope ratio MS was used to analyse the samples for 2H and 18O enrichments. Mean breast milk intake based on saliva samples was significantly higher than that based on urine samples (854·5 v. 812·8 g/d, P = 0·029). This can be attributed to slightly higher isotope enrichments in saliva and to a poorer model fit for urine samples as indicated by a higher square root of the mean square error (14·6 v. 10·4 mg/kg, P = 0·001). Maternal energy expenditure was not correlated with breast milk output. Our study suggests that saliva sampling generates slightly higher estimates of breast milk intake and is more precise as compared with urine and that maternal energy expenditure does not influence breast milk output.