The pharmacokinetic parameters determining antibiotic efficacy are peak concentrations (C_max), minimum (trough) concentrations (C_min), and area under the concentration-time curve (AUC). There is general agreement about the importance of C_max and AUC for aminoglycosides, but this is not so for maintenance of C_min. With in vitro exposures modelling in vivo administration, Pseudomonas aeruginosa reference strain ATCC 27853 (MIC, 1 mg/liter) and a higher-MIC (relatively resistant) clinical isolate (MIC, 4 mg/liter) were used to explore bacteriostatic and bactericidal outcomes. With P. aeruginosa ATCC 27853, kill followed a complete bolus profile with a 30-min postdistribution peak (C_peak30) of 10 mg/liter. The clinical isolate required a C_peak30 bolus profile of 20 mg/liter for kill, and there was no difference between the efficacies of the bolus and infusion exposures. Bolus profiles that were truncated at 8.5 h and producing sublethal effects were then combined with a wide range of C_mins. With a C_peak30 profile of 8 mg/liter, P. aeruginosa ATCC 27853 showed a graded bacteriostatic response until a C_min of 0.8 mg/liter, when complete kill resulted. In contrast, bactericidal effects on the clinical isolate required a C_peak30 profile of 18 mg/liter with a C_min of 1.0 mg/liter. Therefore, C_min also contributes to the bactericidal effect of tobramycin, with requirements showing minor variation with change in MIC. Dosing principles for relatively resistant (higher-MIC) organisms are suggested from the data. Relatively higher aminoglycoside doses via infusion regimens are likely to be needed to generate higher peak concentrations and higher AUC values necessary for bactericidal effect in resistant organisms. Maintenance of trough concentrations on the order of 1.0 mg/liter during the interdose interval will tend to guard against the possibility of inadequate peak and AUC exposures for kill.