Antimicrobial combination therapy has a long history. The potential for synergy between two antimicrobial agents has been sought using in vitro techniques such as disk approximation, checkerboard titration, in vitro killing experiments with fi xed drug concentrations, through the use of in vitro pharmacodynamic models, through animal models, and through retrospective and prospective clinical studies. The most widely examined combinations are those that include aminoglycosides, especially for the treatment of serious Pseudomonas aeruginosa infections, and for the management of enterococcal and staphylococcal endocarditis. The in vitro and animal studies of P. aeruginosa support the concept that combinations of aminoglycosides with β-lactams improve effi cacy and outcomes. Some models suggest that the mechanism of improved effi cacy relates less to synergy and more the prevention of the selection of aminoglycoside-resistant mutants. However, human clinical studies to date have failed to demonstrate convincingly that there are better outcomes with combination therapy in terms of effi cacy or the prevention of resistance emergence. For endocarditis, there has been strong evidence accumulated for combination therapy enterococcal endocarditis using in vitro and animal models, although there are no randomised clinical data to confi rm these. Data supporting the use of combination in staphylococcal therapy has been less robust, and Panton-valentine leukocidin producing S. aureusCell-wall active agent + clindamycin or linezolid the role of combination treatment for this indication is now in question. Ultimately, an in vitro or animal model of the pharmacodynamic interaction of drug classes that can be shown to predict clinical outcomes is still required. Such a model does not currently exist.