Serum vancomycin concentrations determined by fluorescence polarization immunoassay (FPIA) with a specific high-performance liquid chromatography (HPLC) method in preterm neonates were compared. Preterm neonates (< 38 weeks gestational age) requiring vancomycin therapy and serum vancomycin concentration monitoring were enrolled. Peak serum vancomycin concentration samples were collected and independently analyzed by FPIA and HPLC. Multivariate and stratified data analysis was done with mean absolute error and mean percent error as dependent variables and independent variables as postconceptional age, postnatal age, gestational age, weight, and duration of therapy to characterize the findings. A total of 15 paired vancomycin concentrations were analyzed from neonates with a mean gestational age of 30 +/- 4 weeks. The mean percentage error of FPIA versus HPLC vancomycin concentrations was 18.1 +/- 11.1% and the mean absolute error was 3.7 +/- 2.0 mg/l. Postconceptional age, weight, and time from initiation of therapy to sample collection were independent variables which best characterized the overestimation of FPIA vancomycin concentrations. The FPIA vancomycin assay method overestimated actual vancomycin concentrations in preterm neonates. Preterm neonates less than 30 weeks postconceptional age, less than 1,200 g body weight, and duration of therapy greater than 48 h prior to concentration determination had the greatest difference in FPIA and HPLC results. Significant error in pharmacokinetic parameter estimations and dosage adjustments is possible when vancomycin concentrations are determined by FPIA.