Antibacterial drugs account for between 3 and 25% of all prescriptions, between 6 and 21% of the total market value of drugs in a single country, and up to 50% of the drug budget in hospitals. Bacterial infection is widely perceived as disease caused by harmful outside agents which can be isolated and tested to select the best drug for treatment. In fact, the need for any treatment and the pros and cons of different drugs are just as debatable as in any other therapeutic area. Moreover, the bacteria which make up the normal flora of the body fulfil important roles, so that the ecological implications of treatment for the individual and for society should be considered in assessing the costs and consequences of antibacterial treatment. In this review we outline the most important issues relating to the treatment of bacterial infection in the community and in the hospital, contrasting information from developed and developing countries where appropriate. We review the existing literature on economic evaluation, but in general most of the literature deals with containing the costs of antibacterial drugs in hospitals, and there are many gaps in the literature on cost-effectiveness of treatment. Consequently there are still extreme variations in medical practice which present a challenge for future evaluation. As the outcomes of antibacterial treatment are apparent in a few weeks or months, this is an ideal field for testing pharmacoeconomic methodology. The desire to overcome medical practice variation through consensus statements should be avoided. Instead we recommend wider application of decision analysis to acknowledge that choices exist for the diagnosis and treatment of bacterial infection and to gather information about the implications of these choices. Much of the existing literature would be improved by a more explicit definition of costs. Direct costs to the health services should be distinguished from non medical costs. Moreover, the analysis should consider whether savings from one budget result in costs to another health service budget, or to the patient (transfer costs). These deficiencies in cost analysis will be relatively easy to correct. Of more concern is the fact that the efficacy of much antibacterial treatment is either totally debatable, or variable, depending on factors such as the type of patient treated or the quality of delivery of treatment.(ABSTRACT TRUNCATED AT 400 WORDS)