Stable isotope breath tests offer a safe, repeatable non-invasive method of measuring fat digestion. They involve the ingestion of a substrate labelled with 13C followed by serial measurements of the 13C: 12C ratio in expired CO2, from which the percentage of the 13C dose recovered (PDR) can be calculated. However PDR depends on the CO2-production rate. Our aim was to compare results obtained using directly measured CO2-production rates with those calculated from two predicted values. Twelve normal healthy children and twenty-four children with cystic fibrosis (CF) (without or with the normal dose of enzyme supplementation) were studied with 1,3-distearyl, 2[carboxyl[13C]octanoyl glycerol. The volume of CO2 produced (litres/min) was measured at rest for 30 min approximately 3 h after substrate ingestion, and the results were converted to mmol/min. For each subject the expected BMR was calculated from the equation of Schofield (1985), based on sex, age, weight and height, and from these values, CO2-production rate was derived. Surface area was calculated and an estimated value of 5 mmol/m2 per min (Shreeve et al. 1970) was used. Using these three CO2-production rates, three different PDR were calculated and compared. In healthy children there was a close concordance between measured and predicted CO2-production rates, but children with CF had a mean measured CO2-production rate 39% higher than normal children. This use of normal data for predicted CO2-production rates in children with CF underestimates cumulative PDR. If direct measurements of CO2-production rate are not available or impossible to perform the Vco2 obtained from the BMR calculated using the equations of Schofield (1985) or Shreeve et al. (1970) can be used in normal children. However, if accurate results for PDR are to be obtained, CO2-production rates should be measured when performing breath tests in conditions where energy expenditure and/or CO2-production rate are not expected to be normal.