The pharmacodynamics of piperacillin/tazobactam and cefepime were evaluated against extended-spectrum beta-lactatnase (ESBL)-producing organisms. Ten thousand patients were simulated based on ESBL minimum inhibitory concentrations (MICs) from our laboratory (N=39) and on pharmacokinetic data from peer-reviewed literature. The desired proportion of the dosing interval that the concentration remains above the MIC (%T > MIC) for the intermittent bolus regimens was >= 40% for piperacillin/tazobactam and >= 60% for cefepime. The desired C-ss/MIC ratio (where C is the concentration at steady state) was >= 2 for all continuous infusion (CI) regimens. MIC50, MIC90, and %S were, respectively, 64/4 mu g/mL, 1024/4 mu g/mL and 33% for piperacillin/tazobactam and 8 mu g/mL, 16 mu g/mL and 0% for cefepime. For piperacillin/tazobactam, 3.375 g every 4 h (q4h) achieved the highest probability of target attainment (43%), followed by 135 g CI (31%). 3.375 g q6h (27%), 4.5 g q8h (17%) and 6.75 g CI (10%). However, for cefepime, 4 g CI had the highest probability of target attainment (7714). followed by 1 g q8h (65%), 2 g q12h (58%), 3 g CI (46%) and 1 g q12h (27%). Although the probabilities of target attainment for cefepime were higher than for piperacillin/tazobactam, neither agent achieved a high probability of target attainment and should not be used routinely for the treatment of ESBL infections. (c) 2005 Elsevier B.V. and the International Society of Chemotherapy. All rights reserved.