The approach of severe sepsis and septic shock has three fundamental steps: early recognition, hemodynamic resuscitation and early empiric antibiotic therapy with focus control if appropriate. Empiric antibiotic therapy involves two decisions: the choice of antibiotics, which depends on the presumed focus and pathogens, and dosing of antibiotics, which should be appropriate along with a correct route and mode of administration [1].In a recent issue of Intensive Care Medicine, Jan De Waele et al. [2] presented data from the Defining Antibiotic Levels in Intensive care unit patients (DALI) study looking at patient characteristics predictive of non-attainment of pharmacokinetic/pharmacodynamic (PK/PD) targets, namely f T [ MIC of the suspected pathogen for at least 50 and 100 % of the dosing interval, in intensive care unit (ICU) patients receiving eight different b-lactams. The DALI project [3] is an international prospective, multicentre, pharmacokinetic point prevalence study involving 68 hospitals, ten countries, with a total of 343 patients, of whom 259 had infection. In the present study, 57 % of patients received penicillins, 27 % carbapenems, and 16 % cephalosporins. Dosing was at the discretion of the treating clinician.In the present study, antibiotics were given for treatment of infection in 75.5 % of patients; however, there was no information on the rate of microbiological documentation or on the bacterial antibiotic susceptibility. The aim of the study was to evaluate target non-attainment neither for the actual infection nor for the microbiological agents, but envisioning an empirical situation where the least susceptible organism was potentially causing the infection. To overcome this weakness, recognized by the authors, they assume that the concentrations obtained in the DALI study were also the concentrations that would be reached during empirical dosing. Accordingly, the authors calculate the non-attainment of PK/PD targets using the highest European Committee on Antimicrobial Susceptibility Testing (EUCAST) MIC90 data breakpoint for the administered antibiotic among all potential pathogens. The rationale for this choice was that empiric antibiotic dose selection is based on the 'worst-case' scenario in terms of bacterial susceptibility.The authors showed that free antibiotic concentrations remained below the least susceptible MIC during 50 and 100 % of the dosing interval in 66 (19.2 %) and 142 (41.4 %) patients, respectively. By multivariate analysis, they found that intermittent infusion (vs. extended or Intensive