The bactericidal efficacy of amikacin, isepamicin and netilmicin was studied against Pseudomonas aeruginosa and Serratia marcescens over a treatment period of 30 h using two one-compartment in-vitro models with differently designed culture compartments. High bacterial inocula were exposed to fluctuating drug concentrations, simulating human serum concentrations (t m -2 h) during clinical treatment. The same daily dose was administered as 1 h infusions given every 8 h or every 24 h, resulting in peak concentrations of 8 and 24 mg/1 for netilmicin, and 24 and 72 mg/1 for amikacin and isepamicin, respectively. Once-daily dosing was more bactericidal during initial treatment in the in-vitro models (P < 0-01) and at least as effective as thrice-daily dosing in preventing bacterial regrowth, despite a prolonged period of subinhibitory drug concentration before administration of the second dose. Lower ratios of peak concentration to MIC were needed to achieve bactericidal activity (> 99-9% reduction of cfu) after 24 h treatment against S. marcescens compared with P. aeruginosa (/> < 0-01). All nine regimens providing peaks of at least four times the MIC were bactericidal against S. marcescens after 24 h exposure. In contrast, a bactericidal effect against P. aeruginosa occurred only during two of six experiments with peaks of four to nine times the MIC. Similar results were obtained in both in-vitro models of infection. These data suggest insufficient intrinsic activity of the aminoglycosides studied for single drug treatment of P. aeruginosa in the absence of host-defence mechanisms.