BACKGROUNDThe value of baseline positron emission tomography (PET) for predicting overall survival (OS) or disease‐free survival (DFS) is unclear in patients with nondistant metastatic (locoregional only) esophageal carcinoma. The authors tested the hypothesis that, in this setting, the number of PET abnormalities (NPA) would correlate with OS and DFS.METHODSThe authors of the current study analyzed patients with localized esophageal carcinoma (Stages II and III) who had a baseline PET and endoscopic ultrasonography (EUS) and were all treated with chemoradiotherapy followed by surgery. The standardized uptake value (SUV) of PET avid lesions were evaluated for: SUV of the primary, NPA, peak SUV, and total SUV. Correlations were performed with baseline EUS results, OS, DFS, and clinical and pathologic response.RESULTSForty‐seven patients who underwent chemoradiotherapy followed by surgery were analyzed. Most patients had clinical Stage III cancer. NPA was significantly associated with OS (Cox model, P = 0.02; log‐rank test, P = 0.04) and DFS (P = 0.04). Patients with NPA > 1 had a death hazard ratio of 4.49 (reference, NPA = 1). In a multivariate analysis, NPA was independently predictive of OS (P = 0.03). Alternatively, SUV of the primary tumor, peak SUV, total SUV, and EUS clinical stage did not correlate with the type of response, OS or DFS.CONCLUSIONSData from the current study suggest that for nondistant metastatic esophageal carcinoma, baseline PET can predict patient outcome. Baseline NPA (> 1), reflecting the regional nodal metastases, is an independent predictor of OS. Baseline PET may become a useful stratification factor in randomized trials and for individualizing therapy. Cancer 2005. © 2005 American Cancer Society.