BACKGROUND: Dried blood spot (DBS) is a practical sampling strategy for pharmacokinetic studies in neonates. The utility of DBS to determine the population pharmacokinetics (pop-PK) of ampicillin, as well as accuracy versus plasma samples, was evaluated. METHODS: An open-label, multicenter, opportunistic, prospective study was conducted in neonates. Ampicillin concentrations from plasma and DBS (CONCPlasma and CONCDBS) were measured by liquid chromatographic tandem mass spectrometry and analyzed using pop-PK and statistical (including transformation) approaches. RESULTS: A total of 29 paired plasma and DBS samples from 18 neonates were analyzed. The median (range) gestational age and postnatal age were 37 (27-41) weeks and 8 (1-26) days, respectively. The geometric mean of CONCDBS to CONCPlasma ratio was 0.56. Correlation analysis demonstrated strong association between CONCPlasma and CONCDBS (r = 0.902, analysis of variance P < 0.001). Using linear regression transformation, the estimated CONCPlasma (eCONCPlasma) was derived using (CONCDBS - 3.223)/0.51. The median bias and geometric mean ratio improved to -11% and 0.88 (Wilcoxon signed-rank test, P < 0.001), respectively, when comparing eCONCPlasma to CONCPlasma. Furthermore, using pop-PK modeling, the median bias (interquartile range) for clearance and individual predicted concentrations improved to 8% (-11 to 50) and -8% (-34 to 11), respectively, when eCONCPlasma was used. CONCLUSIONS: After transformation, DBS sampling accurately predicted ampicillin exposure in neonates.