OBJECTIVE: To assess the feasibility of a quarterly antibiotic cycling program at two community hospitals and to evaluate its safety and impact on antibiotic use, expenditures, and resistance. DESIGN: Nonrandomized, longitudinal cohort study. SETTING: Two community hospitals, one teaching and one non-teaching. PATIENTS: Adult medical and surgical inpatients requiring empiric antibiotic therapy. INTERVENTION: We developed and implemented a treatment protocol for the empiric therapy of common infections. Between July 2000 and June 2002, antibiotics were cycled quarterly; quinolones, beta-lactam-inhibitor combinations, and cephalosporins were used. Protocol adherence, adverse drug events, nosocomial infections, antibiotic use and expenditures, resistance among clinical isolates, and length of stay were assessed during eight quarters. RESULTS: Physicians adhered to the protocol for more than 96% of 2,494 eligible patients. No increases in nosocomial infections or adverse drug events were attributed to the cycling protocol. Antibiotic acquisition costs increased 31%; there was a 14.7% increase in antibiotic use. Length of stay declined by 1 day. Quarterly variability in the prevalence of vancomycin-resistant enterococci and ceftazidime resistance among combined gram-negative organisms were noted. CONCLUSIONS: Implementation of an antibiotic cycling program is feasible in a community hospital setting. No adverse safety concerns were identified. Antibiotic cycling was more expensive, partly due to an increase in antibiotic use to optimize initial empiric therapy. Quarterly antibiogram patterns suggested that antibiotic cycling may have impacted resistance, although the small number of isolates precluded statistical analysis. Further assessment of this approach is necessary to determine its relationship to antimicrobial resistance.