Gram-negative organisms have become increasingly resistant to both beta-lactam antibiotics and fluoroquinolones. Consequently, aminoglycoside antibiotics have undergone a resurgence in use. Because of the known toxicities of aminoglycoside antibiotics, clinicians have avoided their use, unless no other alternatives were extant. Over the past 2 decades, we have learned much about the relationship between aminoglycoside exposure and the likelihood of a good clinical outcome or the occurrence of nephrotoxicity. For example, minimum inhibitory concentration values > or = 2.0 mg/L lead to unacceptably low probabilities of a good clinical outcome, and infrequent administration of doses (i.e., intervals of 24 h and longer intervals for patients with compromised renal function) plays a central role in minimizing the likelihood of toxicity. Using these new insights, we suggest ways of evaluating the dose and schedule of administration of aminoglycosides in empirical therapy to obtain the highest likelihood of an efficacious and nontoxic therapy.