Gram-negative bacteria that produce extended-spectrum- and metallo-beta-lactamases are being discovered at an alarming rate, while the development of new antimicrobial agents has almost ground to a standstill. A body of experience exists with the detection and treatment of extended-spectrum beta-lactamase-producing organisms (Klebsiella pneumoniae and Escherichia coli), suggesting that knowledge of their existence and dissemination might have an impact on therapeutic choices and patient outcomes via targeted empirical antimicrobial selection and infection control practices. It is unclear whether the same mandate exists for the detection of metallo-beta-lactamases. As dictated by local susceptibility patterns, in many settings worldwide empirical therapy in serious nosocomial infections now requires the use of carbapenems alone or in combination with a second antibiotic that is also effective against Gram-negative pathogens; colistin is advocated as the empirical drug of choice in the setting of multidrug-resistant Pseudomonas aeruginosa infections, although high doses of beta-lactams might prove to be effective.