BACKGROUND & AIMS: Esophageal carcinomas have high fatality rates, making chemoprevention agents desirable. We performed a systematic review with meta-analysis of observational studies evaluating the association of aspirin/nonsteroidal anti-inflammatory drug (NSAID) use and esophageal cancer. METHODS: We evaluated the MEDLINE, BIOSIS, and Web of Science electronic databases (1980-2001); manually reviewed the literature; and consulted with experts. Studies were included if they: (1) evaluated exposure to NSAIDs, aspirin, or both; (2) evaluated esophageal cancer; and (3) reported relative risks or odds ratios or provided data for their calculation. Data were independently abstracted by 2 investigators. The primary and sensitivity analyses used both fixed and random-effects models. RESULTS: Nine studies (2 cohort, 7 case control) containing 1813 cancer cases were identified. All primary summary estimates were homogeneous. Statistical pooling showed a protective association between any use of aspirin/NSAID and esophageal cancer (odds ratio [OR] = 0.57; 95% confidence interval [CI], 0.47-0.71). Both intermittent (OR = 0.82; CI, 0.67-0.99) and frequent medication use were protective (OR = 0.54; CI, 0.43-0.67), with greater protection with more frequent use. Stratified by medication type, aspirin use was protective (OR = 0.5; CI, 0.38-0.66), and NSAIDs had a borderline protective association (OR = 0.75; CI, 0.54-1.0). Any use was protective against both esophageal adenocarcinoma (OR = 0.67; CI, 0.51-0.87) and squamous cell carcinoma (OR = 0.58; CI, 0.43-0.78). CONCLUSIONS: Pooled results support a protective association between aspirin and NSAIDs and esophageal cancer (of both histological types) and provide evidence for a dose effect. These findings support evaluating these agents in clinical trials of high-risk patients.