Background Clinical staging modalities for esophageal cancer are inaccurate at determining prognosis, especially in early-stage patients. We performed a retrospective review of patients with esophageal adenocarcinoma imaged by positron emission tomography before surgical resection to determine whether 18[F]-fluorodeoxyglucose uptake predicted overall survival independently of clinical and pathologic stage. Methods The study is a retrospective review of patients with adenocarcinoma of the esophagus treated by surgery. All patients were imaged with computed tomography and positron emission tomography imaging, and most patients had an endoscopic ultrasound. We compared positron emission tomography standardized uptake values (SUV max) with clinical and pathologic stage and survival. Prognostic variables were assessed by log-rank test, and survival by the method of Kaplan and Meier. Results From January 1996 through June 2004, 50 patients meeting study eligibility criteria were analyzed. Median follow-up for surviving patients was 27 months. The median SUV max was 4.5. Stratification of patients by the median SUV max predicted survival. The 3-year survival was 57% for patients with an SUV max greater than 4.5 and 95% for patients with an SUV max of 4.5 or less ( p = 0.02). The survival advantage of the SUV max 4.5 or less group was also seen in clinically early-stage patients (defined as no adenopathy on computed tomography and positron emission tomography, and by endoscopic ultrasound T1–2 N0), as well as in patients with pathologically early-stage disease (T–2 N0). Conclusions In surgically managed esophageal adenocarcinoma patients, SUV max predicts overall survival. Moreover, SUV max identifies patients who have a poor prognosis from a subset of patients that would otherwise be considered to have early-stage disease.