Objectives: Ertapenem and other carbapenems will be used increasingly, as extended-spectrum β-lactamases become more prevalent even among community-acquired pathogens. There is, however, concern that this use will select for resistances to imipenem and meropenem in nosocomial pathogens, notably Pseudomonas aeruginosa, and we investigated the validity of this concern. Methods: Single-step selection experiments were performed by plating P. aeruginosa cultures on to agar containing doubling dilutions of ertapenem. MIC patterns, outer membrane protein profiles and the effects of efflux inhibitors were examined for selected mutants. Results: At 2-8 × MIC, ertapenem selected (i) for OprD- mutants of P. aeruginosa, with cross-resistance only to carbapenems, (ii) for putative efflux types with broader cross-resistance, and also (iii) for various less familiar phenotypes. Efflux mutants were predominantly, but not exclusively, selected from carbenicillin-hypersusceptible strains and OprD- mutants largely from strains with normal levels of carbenicillin susceptibility. Whilst these data indicate potential cross-selectivity, they must be set against the observation that 20% serum raised the ertapenem MICs, and the drug concentrations needed for mutant selection, by over four-fold, reflecting the compound's strong protein binding. Since, following a 1 g intravenous dose the free ertapenem concentration in the serum falls below 4mg/L - corresponding to the lower of two MIC50 estimates - within 4h (17% of the dosage interval) selectivity in vivo should be minimized. Conclusions: Whilst ertapenem can select for P. aeruginosa mutants with cross-resistance to imipenem and ertapenem in vitro, this selectivity should be minimal under clinical conditions.