The influence of five pharmacokinetic indices on therapeutic response was retrospectively studied in 45 adult patients treated with aminoglycosides for bacterial infections. Subjects were treated for a minimum of five days, had culture and sensitivity reports, and had at least one set of steady-state peak and trough serum aminoglycoside concentrations. Serum drug concentrations were determined by enzyme-multiplied immunoassay or by fluorescence polarization assay. Minimum inhibitory concentrations (MICs) for the drugs were determined by microdilution assays. Cure was determined by negative cultures or absence of clinical evidence of infection. Values for five pharmacokinetic indices were determined for each patient: ratio of steady-state peak serum concentration to MIC (Cssmax/MIC); time that the serum concentration remained above the MIC during a 72-hour period (tsupra-MIC(72)); the intensity index for a 72-hour period (II(72)), which is related but not identical to the area under the curve (AUC), reflecting the contributions of Cssmax/MIC and tsupra-MIC(72); time that the serum concentration was greater than four times the MIC during a 72-hour period (tsupra-(4 X MIC)(72)), and the intensity index related but not identical to AUC greater than four times the MIC (II4 X MIC(72)), which reflects the contribution of Cssmax/(4 X MIC) and tsupra-(4 X MIC)(72). Statistical analysis revealed significant correlations between each of the five indices and the patients' therapeutic responses. The following index values were associated with cures: Cssmax/MIC greater than 4 (and ideally greater than 8); tsupra-MIC(72) of at least 40 hours; tsupra-(4 X MIC)(72) of at least 10 hours; (4) II(72) greater than 400; and II4 X MIC(72) greater than 50. All five pharmacokinetic indices were good predictors of patient outcome. The ratio of maximum steady-state serum aminoglycoside concentration to minimum inhibitory concentration is the index most easily monitored and interpreted.