Objective: To assess the cost-effectiveness of the HLA-B∗15:02 screening policy for the treatment of epilepsy in Hong Kong. Methods: From all public hospitals and clinics in Hong Kong, 13,231 patients with epilepsy who started their first antiepileptic drug (AED) between September 16, 2005, and September 15, 2011 (3 years before and 3 years after policy implementation on September 16, 2015), were identified retrospectively. A decision tree model was constructed to incorporate the real-world data on AED prescription patterns, incidences of AED-induced Stevens-Johnson syndrome (SJS)/toxic epidermal necrolysis (TEN), costs of AED treatments, SJS/TEN treatment, and HLA-B∗15:02 testing, and quality of life. Cost-effectiveness of the screening policy was analyzed for 3 scenarios: (1) current real-world situation, (2) "ideal" situation assuming full policy adherence and preferable testing practices, and (3) "extended" situation simulating the extension of HLA-B∗15:02 screening to phenytoin in ideal practice. Results: The current screening policy was associated with an incremental cost-effectiveness ratio of US $85,697 per quality-adjusted life year (QALY) compared with no screening. The incremental cost-effectiveness ratio was estimated to be US $11,090/QALY in the ideal situation and US $197,158/QALY in the extended situation. Conclusions: The HLA-B∗15:02 screening policy, as currently practiced, is not cost-effective. Its cost-effectiveness may be improved by enhancing policy adherence and by low-cost point-of-care genotyping. Extending the screening to phenytoin would not be cost-effective because of the low incidence of phenytoin-SJS/TEN among HLA-B∗15:02 carriers.