Background The broad antibacterial spectrum of piperacillin/tazobactam makes the combination suitable for the treatment of nosocomial bacterial central nervous system (CNS) infections. As limited data are available regarding piperacillin CNS exposure in patients without or with low-grade inflammation, a clinical study was conducted (1) to quantify CNS exposure of piperacillin by cerebral microdialysis and (2) to evaluate different dosing regimens in order to improve probability of target attainment (PTA) in brain. Methods Ten acute hemorrhagic stroke patients (subarachnoid hemorrhage, n = 6; intracerebral hemorrhage, n = 4) undergoing multimodality neuromonitoring received 4 g piperacillin/0.5 g tazobactam every 8 h by 30-min infusions for the management of healthcare-associated pneumonia. Cerebral microdialysis was performed as part of the clinical neuromonitoring routine, and brain interstitial fluid samples were retrospectively analyzed for piperacillin concentrations after the first and after multiple doses for at least 5 days and quantified by high-performance liquid chromatography. Population pharmacokinetic modeling and Monte Carlo simulations with various doses and types of infusions were performed to predict exposure. A T->MIC of 50% was selected as pharmacokinetic/pharmacodynamic target parameter. Results Median peak concentrations of unbound piperacillin in brain interstitial space fluid were 1.16 (range 0.08-3.59) and 2.78 (range 0.47-7.53) mg/L after the first dose and multiple doses, respectively. A one-compartment model with a transit compartment and a lag time (for the first dose) between systemic and brain exposure was appropriate to describe the brain concentrations. Bootstrap median estimates of the parameters were: transfer rate from plasma to brain (0.32 h(-1)), transfer rate from brain to plasma (7.31 h(-1)), and lag time [2.70 h (coefficient of variation 19.7%)]. The simulations suggested that PTA would exceed 90% for minimum inhibitory concentrations (MICs) up to 0.5 mg/L and 1 mg/L at a dose of 12-16 and 24 g/day, respectively, regardless of type of infusion. For higher MICs, PTA dropped significantly. Conclusion Limited CNS exposure of piperacillin might be an obstacle in treating patients without general meningeal inflammation except for infections with highly susceptible pathogens. Brain exposure of piperacillin did not improve significantly with a prolongation of infusions.