OBJECTIVES: Management of patients presenting with influenza-like-illness (ILI) is a source of cost and clinical concern for hospital systems. Influenza test methods vary in cost, accuracy, and convenience.  We compared the clinical and economic impact of different test modalities, such as: rapid antigen followed by provider judgment vs. Roche’s point of care polymerase chain reaction (PCR) test. METHODS: A budget impact model was developed to compare rapid antigen testing followed by provider judgment to resolve negative results versus the cobas® Influenza A/B nucleic acid test used on the cobas® Liat® System (cobas® Liat® test) in patients presenting with ILI to a healthcare system. The model is based on 100,000 patients stratified by age and risk status presenting to the hospital or physician office. Based on diagnosis, patients receive an antiviral, antibiotic, or no treatment and are sent home/discharged, admitted, or placed into observation. Complications or an additional episode of care are also assessed. Patient management and cost for both strategies was then compared. Inputs were derived from published literature, HCUP data analysis, and clinical expert opinion. RESULTS: The cobas® Liat® test brings the high sensitivity of PCR to the point of care, enabling lab quality results at the time of visit.  Compared to rapid antigen testing with provider judgment, Roche’s cobas® Liat® test increased appropriate therapy selection (50% vs. 25%) while decreasing inappropriate therapy selection (3% vs. 54%). Appropriate therapy defined as antiviral use in influenza positive patients and antibiotic use in influenza negative patients. The cobas® Liat® test reduced resource utilization (admissions, complications, observations) in approximately 1,633 patients and overall cost of patient management was reduced by $3,408,853. CONCLUSIONS: Adoption of Roche’s cobas® Liat® test may improve appropriate therapy use and reduce resource utilization resulting in overall patient management cost savings. Further validation using real world data is warranted.