Four premature neonates and eight infants 1-19 months old received caffeine for apnea. The usual morning oral dose was substituted by 1,3,7 13C-trimethylxanthine (13C-tri CAF) as the citrate salt. Five breath samples were collected the day before (day 1) and the day of 13C-tri CAF administration (day 2). Plasma (after each breath collection) and urine were collected on day 2. 13C-CO2 exhalation was determined by isotope ratio mass spectrometry. Caffeine and its metabolites were measured using high-pressure liquid chromatography. Assessment of the labeled CO2 in the breath revealed no detectable 13C-tri CAF N-demethylation activity in infants before 45 wk postconceptional age. However, demethylation (as urinary metabolites) has been detected before that age. Two-, 4-, and 6-h cumulative excretion of 13C-tri CAF as 13C-CO2 increased with postnatal age and correlated with caffeine plasma clearance (r = 0.840, p less than 0.01). These results were consistent with those obtained for urinary metabolites. In one infant (19 months old) the cumulative excretion of 13C-CO2 while crying was 65% of the value observed during quiet breathing. The measurement of caffeine demethylation using the caffeine CO2 breath test is feasible in infants and is a safe and noninvasive method to determine age related changes in P4501-dependent N-demethylase activity.