Variations in background 2H and 18O abundances in body water influence the accuracy and precision of the 2H218O method for determination of energy expenditure. To investigate the effect of weaning during infancy on background 2H and 18O abundances, urine samples from 44 breast- or formula-fed infants aged 5-16 wk were analyzed. 2H and 18O abundances were significantly higher (P less than 0.001) in breast- than in formula-fed infants. The relationship between 2H and 18O abundances was linear and independent of diet [slope, 4.16 +/- 0.43 (SE)]. By use of this information, the effect of weaning on the accuracy of the 2H218O method was evaluated, taking into account the effect of 2H-18O abundances in the isotope loading dose. In infants weaned completely from breast milk to formula during the measurement, energy expenditure can be overestimated by 18.0%, even if 2H-18O abundances in the isotope dose equal the ratio of naturally occurring background changes. However, this error can be reduced to less than 3.0% by manipulating the study duration and isotope dose. During gradual weaning, the overestimation of energy expenditure is only 0.3-2.0%.