The in-vitro susceptibility of an organism and the pharmacokinetics of an antimicrobial agent are two basic factors on which the choice of standardised treatment regimens is based. However, the inter-individual variability of these factors, which modifies the exposure of bacteria to an antibiotic in terms of time and quantity, is not usually taken into account. In 87 patients treated with beta-lactams (ceftriaxone, cefepime or piperacillin), the probability of failure was greater when the infectious process was located in tissues with barriers to the distribution of beta-lactams. Mean MICs of piperacillin and cefepime, but not ceftriaxone, were below the breakpoints in cases of both recovery and failure, but organisms isolated from patients with a poor outcome had higher MICs. Therefore, the use of breakpoints to determine the susceptibility of microorganisms was not satisfactory in predicting the outcome for a large number of patients. If MICs are determined and plasma concentrations are monitored, dosages can be adjusted according to these parameters, thereby allowing antibiotic treatment to be individualised.