BACKGROUND The management and clinical outcome of patients suffering esophageal trauma depends on a prompt diagnosis. The detection of esophageal injuries by clinical examination, esophagography, or computed tomography is limited. This study aimed to assess the yield and clinical utility of flexible esophagoscopy (FE) in the diagnosis of traumatic esophageal injuries. METHODS During 7 years, we conducted a retrospective (1998-2003) and prospective (2003-2005) study of 163 victims admitted to a trauma hospital, and submitted to FE because of suspected esophageal trauma. Esophageal injury was defined as laceration or perforation, hematoma, abrasion, hematin spots, or ecchymosis. The endoscopic diagnosis was compared with surgical findings or clinical follow-up. RESULTS No traumatic lesion was observed in 139 patients (85.3%), esophageal injuries were detected in 23 (14.1%), and one examination was inconclusive (esophageal stricture, 0.6%). Lacerations were detected in 14 patients and confirmed surgically. Esophageal contusion was observed in nine patients and out of these, five patients underwent surgical exploration and four were managed nonoperatively. The assessment of esophageal injury by FE demonstrated 95.8% sensitivity, 100% specificity, 99.3% accuracy, 100% positive predictive value, and 99.2% negative predictive value. The likelihood ratio for a negative examination was 0.041 and the Youden J Index was 99.2%. CONCLUSIONS FE appears to be an accurate diagnostic tool in the assessment of esophageal injuries. Two main lesions were noted: laceration and contusion. Laceration requires surgical repair. Contusion represents a nonperforative injury of the esophageal wall, requires correlation with computed tomography, and may be managed nonoperatively.