Carbapenems have been the most successful beta-lactam antibiotics in evading bacterial resistance. Nevertheless, acquired carbapenemases are increasingly reported, mostly in Pseudomonas and Acinetobacter isolates but occasionally also in Enterobacteriaceae. They include beta-lactamases of classes B (IMP and VIM), D (OXA-23 to -27) and A (IMI, KPC, NMC and SME). Major outbreaks of producers have occurred in a few centers and, in the US, there has been progressive erosion of carbapenem activity against Acinetobacter species, maybe due to carbapenemases. Acquired carbapenemases are still sufficiently rare not to have placed widespread constraints on chemotherapy, but there is reasonable concern that they will become a greater problem in the future. This is a good argument for continued caution with carbapenem use, and for extending this prudence to the oral and long half-life carbapenems shortly to become available. Most carbapenemase producers are broadly resistant to beta-lactams, and many are also resistant to fluoroquinolones and aminoglycosides. Clinicians facing infections caused by carbapenemase producers consequently are forced to use 'unusual' antibiotics such as polymyxins, isepamicin, minocycline and sulbactam (Pfizer Inc), which has inherent activity against A baumannii. Carbapenemase inhibitors might be developed in the future, despite difficulties in choosing the range of enzymes to target and obtaining broad-spectrum inhibition. For now, the best pharmaceutical strategy seems to lie in the development of novel antimicrobial classes with anti-Gram-negative activity, rather than in overcoming carbapenemases directly.