Newborns are often exposed to medication errors in hospitals. Identification and understanding the causes and risk factors associated with medication errors will help to improve the effectiveness of medication. We sought to compare voluntary incident reports and direct observation in the identification of medication errors. We also identified corresponding risk factors in order to establish measures to prevent medication errors. Medication errors identified by a clinical pharmacist and those recorded in our incident reporting system by caregivers were analysed. Main outcomes were rates, type and severity of medication error, and other variables related to medication errors. Ultimately, 383 medication errors were identified by the clinical pharmacist, and two medication errors were declared by caregivers. Prescription errors accounted for 38.4%, preparation errors for 16.2%, and administration errors for 45.4%. The two variables significantly related to the occurrence of medication errors were gestational age < 32.0 weeks (p = 0.04) and the number of drugs prescribed (p < 0.01).Conclusion: Caregivers underreported the true rate of medication errors. Most medication errors were caused by inattention and could have been limited by simplifying the medication process. Risk of medication errors is increased in newborns < 32.0 weeks and increases with the number of drugs prescribed to each patient. What is Known: • Newborns in hospitals are particularly susceptible to medication errors. • Identification and understanding the reasons for medication errors should help us to establish preventive measures to reduce the occurrence of such errors. What is New: • Direct observation of the medication process, though time consuming, is essential to accurately assess the frequency of medication errors, which are underreported by caregivers. Most medication errors are caused by inattention and could be limited by simplifying the medication process. • The risk of medication errors was significantly increased in very preterm newborns (< 32 weeks) and when the number of prescription per patient increased.