OBJECTIVE:: To determine the half-life of serum caffeine concentrations and its relation to apnea of prematurity (AOP) after caffeine is discontinued in preparation for hospital discharge. STUDY DESIGN:: Prospective cohort study involving preterm infants with gestational ages ≤ 33 weeks at birth. After caffeine was discontinued, serum caffeine concentrations and electronic detection of pathologic apnea, defined a priori, were obtained at 24 and 168 h, respectively. RESULT:: Caffeine levels decreased from 13.3 ± 3.8 to 4.3 ± 2 mg l (n = 50, mean ± s.d.) at 24 and 168 h, respectively (P < 0.01). The mean caffeine half-life was 87 ± 25 h at 35 ± 1 weeks postmenstrual age. Seven days after discontinuation of caffeine, 64% of the infants had pathologic apnea. CONCLUSION:: Hospital discharge planning for preterm infants with a history of AOP should be carefully considered after discontinuing caffeine. This study showed that caffeine may not reach subtherapeutic levels until around 11-12 days.