Barrett's esophagus is a well-recognized precursor of esophageal adenocarcinoma. Surveillance of Barrett's esophagus patients is recommended to detect high-grade dysplasia (HGD) or early cancer. Because of wide variation in the published cancer incidence in Barrett's esophagus, the authors undertook a systematic review and meta-analysis of cancer and HGD incidence in Barrett's esophagus. Ovid Medline (Ovid Technologies, Inc., New York, New York) and EMBASE (Elsevier, Amsterdam, the Netherlands) databases were searched for papers published between 1950 and 2006 that reported the cancer/HGD risk in Barrett's esophagus. Where possible, early incident cancers/HGD were excluded, as were patients with HGD at baseline. Forty-seven studies were included in the main analysis, and the pooled estimate for cancer incidence in Barrett's esophagus was 6.1/1,000 person-years, 5.3/1,000 person-years when early incident cancers were excluded, and 4.1/1,000 person-years when both early incident cancer and HGD at baseline were excluded. Corresponding figures for combined HGD/cancer incidence were 10.0 person-years, 9.3 person-years, and 9.1/1,000 person-years. Compared with women, men progressed to cancer at twice the rate. Cancer or HGD/cancer incidences were lower when only high-quality studies were analyzed (3.9/1,000 person-years and 7.7/1,000 person-years, respectively). The pooled estimates of cancer and HGD incidence were low, suggesting that the cost-effectiveness of surveillance is questionable unless it can be targeted to those with the highest cancer risk.