PURPOSE: To drive decisions on antibiotic therapy in the intensive care unit (ICU), we developed an antibiotic care bundle (ABC-Bundle) with evidence-based recommendations (EBRs) for antibiotic prescriptions. METHODS: We conducted a three-step prospective study. First, a systematic review was performed of the literature reporting EBRs for antibiotic usage in the ICU. Second, we developed an ABC-Bundle through a two-round, RAND-modified Delphi method with an international expert panel, including the most relevant EBRs on a 9-point Likert scale. Those EBRs that were considered mandatory by >50% of the experts were included in the bundle. Third, we assessed the adherence to and applicability of the bundle in two mixed university ICUs. RESULTS: Out of 1190 potentially relevant articles, 14 (four guidelines, four randomised controlled trials and six systematic reviews) fulfilled the eligibility criteria. Six EBRs were classified as relevant: 1. Provide rationale for antibiotic start; 2. Perform appropriate microbiological sampling; 3. Prescribe empirical antibiotic therapy according to guidelines (Day 1); 4. Review diagnosis; 5. Evaluate de-escalation based on microbiological results (Days 2-5); and 6. Consider discontinuation of treatment (Days 3-5). Daily adherence to the ABC-Bundle, prospectively assessed in 861 days of therapy in 142 ICU patients, ranged from 2% to 37%. CONCLUSION: The ABC-Bundle is a novel tool to improve delivery of appropriate antibiotic therapy to ICU patients. The low adherence in the prospective cohorts confirms the significant role that the ABC-Bundle could play in an antibiotic stewardship programme in the ICU setting.