The rationale for pharmacokinetic monitoring of aminoglycosides is critically reviewed. Retrospective studies suggest that for optimal antibacterial effect, peak serum gentamicin concentrations should exceed 5 μg/ml, despite the fact that these concentrations are indirect measures of the concentration of drug at the site of infection. When quantitative results of antimicrobial susceptibility are known (e.g. MIC or MBC), limited data suggest that for most infections, the peak serum aminoglycoside concentration should exceed the minimum inhibitory concentration by four‐fold. However, the optimal duration for which the serum concentration should exceed the MIC or MBC during each dosing interval and the detrimental effect of prolonged subinhibitory drug concentrations have not been evaluated. Furthermore, the immunological competence of the host and the pathogenicity of the infecting organism are important factors in achieving antibacterial response.Using serum creatinine as an indirect and relatively late indicator of nephrotoxicity, nadir gentamicin concentrations greater than 2 μg/ml may predispose patients to develop nephrotoxicity. In addition, recent information indicates that patients who accumulate excessive amounts of aminoglycosides In their tissues may be at higher risk for developing nephrotoxicity; these patients may be identified based on the extent of accumulation in their nadir concentrations with continuous dosing. The aminoglycosides diffuse into the inner ear fluids slowly and diffuse out with a half‐life of decline in inner ear fluid concentrations slower than that in serum. High transient peak serum concentrations probably do not contribute significantly to the risk of ototoxicity. However, there is evidence from early clinical trials, studies using continuous infusions of aminoglycosides, and animal studies indicating that elevated nadir serum concentrations relate to the development of ototoxicity.There is considerable interpatient variability in the peak serum concentration, even when identical dosages based on body weight or surface area are administered. Similarly, the half‐life for decline In serum concentrations is highly variable from patient to patient, even in patients with stable normal renal function. Absorption after intramuscular administration is reliable in most patients, although critically ill patients may experience erratic absorption. The distribution of aminoglycosides is altered in obese patients because of differences in extracellular fluid content between fat and other tissues. Maintaining serum aminoglycoside concentrations in the desired therapeutic range is difficult in patients with impaired renal function, and frequent monitoring of serum concentrations is essential in these patients. Neonates, infants and children differ from adults in their ability to eliminate aminoglycosides and in their central volume of distribution. Fever, state of hydration, “third‐spaces”, and concurrent penicillin therapy may also affect the pharmacokinetics of the aminoglycosides. Diseases which affect some of the factors mentioned above (e.g. burns) may cause unpredictable changes in pharmacokinetic behavior.The approach at St. Jude Children's Research Hospital to pharmacokinetic monitoring of aminoglycoside therapy in children with cancer is presented. This method accounts for tissue accumulation, and allows for dosage adjustments using individualized pharmacok