Therapeutic drug monitoring (TDM) is generally unnecessary in caffeine therapy for apnea of prematurity because the normal blood-caffeine concentrations of preterm infants are markedly lower than those at which caffeine intoxication occurs. However, several reports have mentioned preterm infants developing toxicity. In this retrospective observational study, conducted at a tertiary center in Kagawa, Japan, we evaluated the correlation between the maintenance dose and blood-caffeine concentrations and determined the maintenance dose leading to toxic caffeine concentrations.. Preterm infants were treated with caffeine citrate for apnea of prematurity between 2018 and 2021. Our primary outcome measure was the maintenance dose leading to toxic caffeine concentrations. Twenty-four preterm infants (gestational age, 27 ± 2.9 weeks; body weight, 991 ± 297 g) were included, and 272 samples were collected for analysis. The caffeine dose and blood-caffeine concentration were positively correlated (p < 0.05, r = 0.72). At doses of ≥ 8 mg/kg/day, 15% (16/109) of patients had a blood-caffeine concentration above the toxic concentration. Patients who receive doses of ≥ 8 mg/kg/day are at risk of reaching the toxic blood-caffeine concentration. Although it is unknown whether toxic caffeine concentrations are detrimental to neurological prognosis, the administration of high-dose (≥ 8 mg/kg/day) caffeine may require TDM to prevent blood-caffeine concentrations from reaching toxic levels.