OBJECTIVES: We evaluated the cost-effectiveness of a policy of universal influenza vaccination across the US population, compared with the current approach of vaccinating selected age and risk groups. METHODS: We modeled costs and outcomes of seasonal influenza under a policy of universal vaccination compared with current policy, taking a societal perspective. The population was stratified to model age-specific (<5, 5-17, 18-49, 50-64, and 65+ years) vaccination coverage assumptions, vaccine efficacy, and risks of influenza complications. Probability of influenza-related illness (ILI) and complications, associated health care utilization, direct and indirect costs, and survival were estimated for each policy. Model inputs were derived from published literature and public sources. No herd effects were considered.  For a season’s intervention, ILI cases in that year and lifetime costs and QALYs lost were calculated for each policy and used to derive incremental cost-effectiveness ratios for the overall US population. One-way and probabilistic sensitivity analyses (PSA) were conducted. RESULTS: In base-case analysis, current policy led to 63 million ILI cases per year, resulting in loss of 857,000 QALYs and costs of $115 billion (both discounted at 3% annually; costs in 2008 USD), while universal vaccination resulted in 61 million cases, 826,000 QALYs lost, and $112 billion in costs. Universal vaccination dominated current recommendation, costing $3 billion less and averting 2 million cases, resulting in 31,000 fewer QALYs lost. Results were most sensitive to the percentage of unvaccinated adults developing ILI and coverage assumptions with universal policy. PSA indicated considerable uncertainty of results, with universal coverage was dominant in 54%, and dominated in 20%, of iterations.  CONCLUSIONS: Our results indicate that a recommendation in of universal vaccination in the US is likely to result in lower costs and improved outcomes compared with current recommendation, and that this likelihood depends on seasonal attack rates and coverage assumptions.