Aminoglycosides are of major importance in treating Pseudomonas aeruginosa pneumonia (PAP). However, their efficacy may be compromised by low-level resistance caused by the inducible MexXY multidrug efflux pump. In the present study, the impact of the MexXY efflux pump was investigated in vivo in an experimental model of PAP in rabbits treated with intravenous tobramycin. Three strains were used to induce PAP in rabbits: PAO1 (wild-type strain; MIC 1 mg/L), mutant 11B (mexX::Tn501; no expression of MexXY; MIC 0.5 mg/L) and mutant MutGR1 (MexZ null; constitutive expression of MexXY; MIC 2 mg/L). Five hours after inoculation, treatment with tobramycin (10 mg/kg) was implemented (peak serum concentration 30 mg/L). The animals were killed humanely 48 h after inoculation, and the residual pulmonary bacterial concentration was determined. Selection of bacteria expressing MexXY was determined by plating lung homogenates on agar plates containing antibiotic. Mean bacterial counts (log(10) CFU/g) for treated vs. untreated rabbits were 6.26 and 8.13 (p < 0.0001), 6.00 and 8.38 (p < 0.001), and 7.25 and 8.79 (p 0.04) for PAO1, 11B and MutGR1, respectively, with an overall mortality rate of 0% vs. 8.9% (p < 0.01). MexXY-overexpressing bacteria were recovered from three (21%) treated rabbits. The C(max)/MIC ratio was the parameter that was best associated with tobramycin efficacy. The bacteria overexpressing MexXY, recovered from lung, occurred with a C(max)/MIC window of 19-26. It was concluded that the experimental PAP model highlights poor tobramycin bacteriological efficacy in vivo, contrasting with survival gain, and that the contribution of the MexXY system to this low level of tobramycin efficacy is modest. Finally, this model appears to be suitable for the investigation of new anti-pseudomonal therapeutic strategies.