ObjectiveTo compare various strategies for antibacterial administration in terms of patient outcomes, overall costs and quality of care provided.DesignProspective, nonblind, randomized, clinical study.SettingTertiary care hospital in Greece from November 1995 to June 1996.Patients and participants458 patients admitted to the internal medicine department who received antibacterial therapy for infectious diseases.MethodsPatients were randomized into 4 different antibacterial administration policies defined by various levels of restriction control. Efficacy and resource use data were obtained from clinical study case report forms, the hospital financial database and physician expert opinion. Outcomes included complete infection control, disease improvement, unchanged patient condition, infection needing surgical treatment, and death. Direct medical costs were estimated. The perspective adopted was that of the healthcare system (hospital budget; third-party payor). Cost-minimisation analysis was based on cost per patient treated.Results382 eligible patient records examined showed no significant difference in clinical outcomes among patient groups. Baseline analysis showed the strict antibacterial control policy to produce statistically significant differences (p < 0.05) in various resource parameters. Accordingly, compared with all other patient groups, total cost per patient for that strategy was reduced by 26 to 30%. Also, patients in that group received fewer drug doses and underwent fewer treatment days, and antibacterial treatment was modified in fewer cases for these patients.ConclusionStrict control of antibacterial administration in this hospital setting achieved lower direct medical costs with no harmful effect on patient outcomes or quality of care provided. Such a policy appears to be a useful option for both physicians and administrators.