OBJECTIVES: To investigate the potential benefit of adding genetic testing to a pharmacist-managed anticoagulation service for hospitalized patients on warfarin. METHODS: Using decision analytic modeling we constructed a decision tree model. The decisions used were a treat-all and a treat-none approach. The outcome of interest was minimizing the number of adverse events (bleeding or thrombitic event). Model parameters were primarily based on institutional data. Additional information from the literature or expert opinion was used if needed. One-way sensitivity analyses were performed on unknown parameters to determine decision thresholds. Two parameters, percentage of patients benefiting from testing and rate of reduction of adverse event rates due to genetic test results, were given special attention as no prior information is available. Probabilistic sensitivity analysis was also conducted. A willingness-to-pay of $35,000 per avoided adverse event was used as the decision threshold. RESULTS: Sensitivity analysis showed the decision choice to be affected by the values of both parameters of interest (20% prevalence and 8% reduction respectively). Baseline analysis resulted in the treat-all approach to cost an additional $102 per patient on average and avoid 8 adverse events per 10,000 patients. Probabilistic analysis determined the point at which each decision was equally likely to be optimal was at a willingness-to-pay of $140,000. Holding the reduction rate constant the rate of patients benefitting from the test would need to be as high as 40% before the treat-all approach becomes optimal. Similarly, the reduction rate would need to be as high as 16% for the treat-all approach to be optimal. CONCLUSIONS: The percentage of patients that would benefit from testing, as well as the reduction in adverse event rates due to testing, influence the decision of treatment approach. The addition of testing all patients could potentially be cost-effective with sufficiently high enough prevalence and adverse event reduction rates.