Twenty-three premature infants receiving caffeine maintenance therapy were followed prospectively for several months. Three to nine determinations of caffeine half-life (peak and trough caffeine levels) were made in each baby. This first longitudinal study confirmed that the half-life of caffeine is prolonged during the neonatal period (97.6 + 32 hours and for as many as 38 weeks' gestation in several very premature babies). Contrary to previous assumptions, gestational age and postconceptional age seem to be closely related to maturation of hepatic caffeine elimination after the neonatal period, although a high variability of caffeine half-life was observed between infants. Adult values (6 hours) were obtained about 60 weeks postconceptional. Caffeine half-life was greatly increased in two infants who had cholestatic hepatitis secondary to prolonged parenteral alimentation and one infant who was breast-fed exclusively. In this last case, the role of maternal hormones in repressing the normal enzymatic maturation process is strongly suspected. Adequate blood levels of caffeine were usually obtained with a caffeine half-life greater than 30 hours up to 46 weeks postconceptional with a dose of 5 mg/kg of caffeine citrate. Caffeine predose monitoring is adequate up to 46 weeks postconceptional, and caffeine half-life determination is mandatory whenever the trough level is too high or too low, icterus is present, and from 46 to 50 weeks postconception.