A comprehensive, multiphasic review of gentamicin and tobramycin utilization was undertaken with audits of the microbiological sensitivity of Gram-negative pathogens; indications for the prescription of aminoglycoside agents; the utilization of assay services; the adequacy of clinical drug delivery by measures of serum antibiotic levels; and the assessment of adverse outcomes by markers of nephrotoxicity. The great majority of clinical isolates of target organisms (n = 4208) was more sensitive to gentamicin (96%) and to tobramycin (99%) than to all alternative agents, including first- and third-generation cephalosporin agents. A review of the indications for the prescription of aminoglycoside agents by clinical criteria showed that in 85.6% of 278 documented cases, the choice of agent was appropriate by clinical and microbiological criteria. In a substantial (77.6%) proportion of the 511 patients who were receiving therapeutic courses of an aminoglycoside agent, serum drug assays had been performed. Assay data could not be interpreted adequately in 52.6% of 3079 assayed cases as a result of inadequate data on administration regimens (39.7%) or sampling regimens (12.9%). Where sampling was documented adequately, there was extreme variation (zero to five hours) in post-dose sampling. In only 33.2% of cases could it be concluded unambiguously that the patients were receiving safer, adequate therapy for clinically significant infections, 5.6% of patients were receiving potentially toxic doses, and 8.6% of patients showed suboptimal concentration profiles. The majority of potentially toxic levels were associated with adverse effects.(ABSTRACT TRUNCATED AT 250 WORDS)