Sir, Sauermann et al. recently investigated the influence of human cerebrospinal fluid (CSF) on the antimicrobial activity of fosfomycin in vitro. Bacterial time–kill curves were performed in different media using fosfomycin concentrations below and above the MIC for two Staphylococcus aureus isolates with high and low susceptibility according to BSAC and CLSI definitions. In addition, they exposed the test strains to the fosfomycin pharmacokinetic profile measured at steady-state in the CSF of a selected patient with external ventricular drain-associated infection from a study published previously by our group in CSF obtained from pooled samples. Despite initial killing, the in vitro simulation of the selected pharmacokinetic profile showed terminal regrowth of the S. aureus isolates. The authors deduce from their experiments that the antibacterial activity of fosfomycin is notably reduced in human CSF and may not be sufficient to achieve bactericidal effects in the CNS. This is in contrast to the results of a previous in vivo pharmacokinetic study by our group, indicating that intravenous fosfomycin should provide sufficient antimicrobial concentrations in the CSF and, therefore, might qualify for the treatment of nosocomial intracranial infections. Though we cannot definitely rule out that in our study the combination of fosfomycin with other antibiotics as recommended by experimental and clinical data contributed to enhanced bacterial killing by fosfomycin with complete resolution of intracranial infection in all our patients, concerns arise when extrapolating the presented in vitro data to real-life clinical conditions. Unfortunately, Sauermann et al. do not give further information on the biochemical and cellular composition of the used CSF samples or the clinical condition of the respective patients from whom the CSF was obtained. This would have been of paramount interest for the correct interpretation of their in vitro results. It has to be emphasized that critical illness not only grossly alters the pharmacokinetics, but also influences the pharmacodynamics of anti-infective agents by a significant amount. In the case of nosocomial intracranial infections or intraventricular haemorrhage, the composition of CSF is significantly altered. CSF biochemistry and cytology reflect the nature and severity of the infection with elevated white cell count and reduced CSF glucose. Anaerobic metabolism of glucose results in lactate production, which interferes with CSF acidity. Lysis of cells releases proteins and other intracellular constituents, such as phosphorylated carbohydrates. As stated by Sauermann et al., glucose-6phosphate (G-6-P) is not detectable in human CSF under physiological conditions. Both aspects, namely CSF acidity and the release of phosphorylated carbohydrates, are of importance regarding the antimicrobial activity of fosfomycin: (i) fosfomycin activity is dependent on pH and its microbial killing is impaired at higher pH; and (ii) further, this substance utilizes the hexose phosphate transport system, which is inducible by G-6-P, as a way of entry into microorganisms. In our opinion, the ready availability of G-6-P can be regarded as an important biological determinant for the efficacy of fosfomycin in vivo. However, this assumption requires empirical proof. In our in vivo pharmacokinetic study, all patients suffered from haemorrhagic contamination of ventricular CSF due to primary disease and presented with ventriculitis. In view of the in vitro results presented by Sauermann et al. and our in vivo observations, we have reason to assume that the altered biochemical characteristics of the haemorrhagic and inflamed CSF seen in our patients indeed enhanced the antimicrobial activity exerted by fosfomycin under pathological conditions in human CSF. Based on these data, we are still convinced that co-administration of fosfomycin could represent an appropriate treatment option for nosocomial device-related intracranial infections with susceptible S. aureus isolates, especially in patients with allergy or intolerance to first-line therapy. Importantly, however, with the continuing emergence of multidrug-resistant staphylococci as frequent causes of nosocomial ventriculomeningitis in the neurocritical care patient population, treatment with novel antistaphylococcal agents with adequate CSF penetration, such as the oxazolidinone linezolid, should be considered.