Background Endoscopic surveillance for non-dysplastic Barrett's esophagus (BE) is contentious and its cost effectiveness unclear. Objective To perform an economic analysis of endoscopic surveillance strategies. Design Cost-utility analysis by using a simulation Markov model to synthesize evidence from large epidemiologic studies and clinical data for surveillance, based on international guidelines, applied in a coordinator-managed surveillance program. Setting Tertiary care hospital, South Australia. Patients A total of 2040 patient-years of follow-up. Intervention (1) No surveillance, (2) 2-yearly endoscopic surveillance of patients with non-dysplastic BE and 6-monthly surveillance of patients with low-grade dysplasia, (3) a hypothetical strategy of biomarker-modified surveillance. Main Outcome Measurements U.S. cost per quality-adjusted life year (QALY) ratios. Results Compared with no surveillance, surveillance produced an estimated incremental cost per QALY ratio of $60,858. This was reduced to $38,307 when surveillance practice was modified by a hypothetical biomarker-based strategy. Sensitivity analyses indicated that the likelihood that surveillance alone was cost-effective compared with no surveillance was 16.0% and 60.6% if a hypothetical biomarker-based strategy was added to surveillance, at an acceptability threshold of $100,000 per QALY gained. Limitations Treatment options for BE that overlap those for symptomatic GERD were omitted. Conclusion By using best available estimates of the malignant potential of BE, endoscopic surveillance of patients with non-dysplastic BE is unlikely to be cost-effective for the majority of patients and depends heavily on progression rates between dysplasia grades. However, strategies that modify surveillance according to cancer risk might be cost-effective, provided that high-risk individuals can be identified and prioritized for surveillance.