OBJECTIVE: Screening for varices is recommended in patients with cirrhosis to institute primary prophylaxis to prevent variceal bleeding. Our aim was to compare the cost-effectiveness of four strategies, including no screening/no prophylaxis, universal screening and primary prophylaxis with β-blockers, universal screening and primary prophylaxis with variceal ligation, and universal institution of primary prophylaxis with β-blockers without screening. METHODS: We constructed a Markov simulation model in two hypothetical cohorts of 50-yr-old patients with cirrhosis (one compensated and one decompensated), who were followed for 5 yr. Transition probabilities were derived from the medical literature, and costs reflected Medicare reimbursement rates at our institution. RESULTS: In patients with compensated cirrhosis, screening and primary prophylaxis with β-blockers is associated with an incremental cost-effectiveness ratio of $3605 per year of life saved. The results were most sensitive to the prevalence of varices and risk of variceal bleeding. In patients with decompensated liver disease, primary prophylaxis without screening was associated with an incremental cost-effectiveness ratio of $1154 per year of life saved. The results were most sensitive to the cost of β-blockers and endoscopy. CONCLUSIONS: Screening for varices is an affordable strategy in compensated liver disease, whereas universal primary prophylaxis with β-blockers is cost-effective in decompensated patients. © 2002 by Am. Coll. of Gastroenterology.