The rationale for appropriate antimicrobial treatment of deep-seated infections has progressively changed over the last 10 years. Although the antimicrobial choice used to be essentially based only on the spectrum of activity and on the in vitro susceptibility of bacterial pathogens, it has become evident that at least two additional factors have to be borne in mind whenever bacteria are located within organs. In the presence of deep-seated infections, it is mandatory to achieve optimal pharmacodynamic exposure with antimicrobials not only in plasma but also at the infection site. This means that both the pharmacokinetic characteristics of the drugs and the anatomical characteristics of the infection sites must be taken into account in order to avoid drug underexposure in the various organs. Additionally, in critically ill patients, particular attention should be given to the possibility that the pathophysiological changes that frequently occur in this setting may significantly alter drug exposure at the infection site, especially for hydrophilic antimicrobials. Greater understanding of the type of antibacterial activity has emphasized the need to choose different dosages and dosing regimens for the various antibacterials. As a general rule, for time-dependent antimicrobials, (i.e. b-lactams, glycopeptides and oxazolidinones), administration of the total daily dosage in multiple doses or as a continuous infusion may be worthwhile, with the intent of favouring more sustained concentrations and thus maximizing the time during which the concentration persists above the minimum inhibitory concentration (T >MIC). Accordingly, in daily clinical practice, optimization of drug exposure with time-dependent antimicrobials may be pursued through maintenance of minimum (trough) plasma trough concentrations (Cmin) above the MIC (Cmin >MIC). Conversely, for concentration-dependent agents (i.e.fluoroquinolones and aminoglycosides), once-daily dosing should be chosen whenever feasible, with the intent of maximizing plasma drug exposure either in terms of the maximum plasma concentration (Cmax) [a Cmax/MIC ratio >10] or in terms of the total daily exposure over time (an area under the plasma concentrationtime curve [AUC]/MIC ratio >100). In fact, for these agents, unlike the time-dependent antimicrobials, higher concentrations mean more rapid and potent bacterial killing. In this context, plasma concentrations are usually considered to be valuable surrogate markers of drug exposure in tissue during deep-seated infections. In fact, it is usually expected that an equilibrium of drug concentrations between plasma and the tissue interstitium – namely, the site where bacterial pathogens aremost frequently located within organs – is achieved. However, it has been shown that plasma concentrations may not always correctly predict concentrations in the tissue interstitium. Drug penetration may be altered in the presence of peculiar pathophysiological conditions as, for example, has been observed in patients with septic shock. In these cases, dose adjustments that are exclusively based on plasma concentration data may therefore be misleading. Interestingly, direct measurement of drug concentrations at the infection site is gaining increasing relevance for appropriate treatment of deep-seated infections with antimicrobials, especially for sites where direct collection of extracellular fluids is feasible, e.g. by direct sampling of epithelial lining fluid in cases of pneumonia, cerebrospinal fluid in cases of meningitis or vitreous humour in cases of endophthalmitis. In some other tissue sites where direct sampling is not feasible, interstitial fluids may be collected by means of microdialysis, which is a probe-based sampling method employed to measure drug concentrations in selected tissues; for example, in the interstitium of muscle and the subcutis in the case of skin and softtissue infections. Conversely, whenever assessment of the extracellular drug concentration in tissues is not feasible either by direct sampling or by microdialysis, measurement is usually performed in tissue homogenates. Indeed, the relevance of findings from studies COMMENTARY Clin Pharmacokinet 2009; 48 (2): 125-127 0312-5963/09/0002-0125/$49.95/0