The use of bioelectrical impedance (BI) analysis as a non-invasive approach for individualizing gentamicin therapy in newborn infants has been investigated in a two phase study. In Phase I, 1/impedance and length were identified as statistically significant predictors of the distribution volume of gentamicin (Adj R2 = 0.78, CV = 12.42%), and length/impedance and post-conceptual age were predictors of total systemic clearance (Adj R2 = 0.83, CV = 14.5%), following the administration of 2.5 mg.kg-1 gentamicin to 17 neonates (gestational age (GA) 27 to 36 weeks). In a prospective validation of these relationships in an independent (Phase II) group of 27 infants (GA 26 to 41 weeks), predicted serum gentamicin concentrations were close to those achieved. Several instances of high prediction errors (predicted minus achieved levels) were observed in infants with known or suspected renal impairment and they caused significant (P < 0.05) perturbation in the bias and accuracy of the models. Daily BI measures over a four to five day period were able to detect individual changes in the fat-free body compartments, which were translated into alterations in gentamicin regimens. This simple, non-invasive and relatively inexpensive bedside technique provides a potentially valuable means to individualize gentamicin therapy without relying on the measurement of serum gentamicin concentration.