This study assessed the rate of errors in intravenous medicine preparation at the bedside in neonatal intensive care units vs the preparation error rate in a hospital pharmacy service. We conducted a prospective observational study between June and September 2013. Ten Spanish neonatal intensive care units and one hospital pharmacy service participated in the study. Two types of preparation errors were considered: calculation errors and accuracy errors. A total of 522 samples were collected: 238 of vancomycin, 139 of gentamicin, 39 of phenobarbital and 88 of caffeine citrate preparations. Of these, 444 samples were collected by nurses in neonatal intensive care units, and 60 were provided by the hospital pharmacy service. Overall, 18 samples were excluded from the analysis. We detected calculation errors in 6/444 (1.35%) and accuracy errors in 243/444 (54.7%) samples from the neonatal intensive care units. In contrast, in samples from the hospital pharmacy service, no calculation errors were detected, but there were accuracy errors in 23/60 (38.3%) samples. CONCLUSIONS While calculation errors can be eliminated using protocols based on standard drug concentrations, accuracy error rates depend on several variables that affect both neonatal intensive care units and hospital pharmacy services. BACKGROUND Medication use is associated with a risk of errors and adverse events. Medication errors are more frequent and have more severe consequences in paediatric patients. Lack of knowledge of drug pharmacokinetics and pharmacodynamics in relation to physiological immaturity makes neonates more vulnerable to medication errors. BACKGROUND Calculation errors are avoided using concentration standard preparation protocols. Accuracy in the preparation process depends mainly on the degree to which commercial drug preparations meet current legal requirements and the syringes and preparation techniques used.