The serum levels of the three macrolides - roxithromycin, clarithromycin and azithromycin - vary considerably. The prediction of the antibacterial effect against extracellular pathogens is based on circulating concentrations of free drug, peak and trough levels, the rate of killing, and the presence of a post-antibiotic effect. Intracellular activity depends on the distribution of the antibiotic and the localization of the bacteria, and is variable. Roxithromycin uptake is greater than that of erythromycin. The intracellular half-life may be long for some compounds (azithromycin>roxithromycin). The intracellular distribution is bimodal, both in the lysosomes and the cytoplasm, but the mechanisms of uptake have not yet been established. At low pH, accumulation is low and macrolides are less active in an acidic medium. Intracellular concentrations cannot readily be predicted on the basis of extracellular levels. Different models have shown that the greater the intracellular concentration, the better the clinical effect. In addition, the transport of macrolides by cells into the infected focus may play an important role in the therapeutic outcome. These factors influence the clinical indications for macrolides, their dosing regimens and breakpoints. In future, macrolides will be developed that are more selective for intracellular infections, while others, which will achieve significant serum levels, will be useful for a broader range of diseases. However, new compounds should be evaluated in different models of infection before clinical studies are instituted, The analysis of failures remains the most important approach in defining concentration/effect relationships.