Background and study aims: Several studies have suggested that a significant minority of esophageal cancers are missed at endoscopy The aim of this study was to estimate the proportion of esophageal cancers missed at endoscopy on a national level, and to investigate the relationship between miss rates and patient and tumor characteristics. Patients and methods: This retrospective, population-based, cohort study identified patients diagnosed with esophageal cancer between April 2011 and March 2012 in England, using two linked databases (National Oesophago-Gastric Cancer Audit and Hospital Episode Statistics). The main outcome was the rate of previous endoscopy within 3-36 months of cancer diagnosis. This was calculated for the overall cohort and by patient characteristics, including tumor site and disease stage. Results: A total of 6943 new cases of esophageal cancer were identified, of which 7.8% (95% confidence interval 7.1-8.4) had undergone endoscopy in the 3-36 months preceding diagnosis. Of patients with stage 0/1 cancers, 34.0% had undergone endoscopy in the 3-36 months before diagnosis compared with 10.0% of stage 2 cancers and 4.5% of stage 3/4 cancers. Of patients with stage 0/1 cancers, 22.1% were diagnosed after ≥3 endoscopies in the previous 3 years. Patients diagnosed with an upper esophageal lesion were more likely to have had an endoscopy in the previous 3-12 months (P=0.040). The most common diagnosis at previous endoscopy was an esophageal ulcer (48.2% of investigations). Conclusion: Esophageal cancer may be missed at endoscopy in up to 7.8% of patients who are subsequently diagnosed with cancer. Endoscopists should make a detailed examination of the whole esophageal mucosa to avoid missing subtle early cancers and lesions in the proximal esophagus. Patients with an esophageal cancer may be misdiagnosed as having a benign esophageal ulcer.