Recent enthusiasm for the cervical esophagogastric anastomosis has arisen because of its perceived low morbidity. Although catastrophic complications of a cervical esophagogastric anastomosis are unusual, they can and do occur, and prevention is possible if the potential for them is recognized. Among 856 patients undergoing a cervical esophagogastric anastomosis after transhiatal esophagectomy, catastrophic cervical infectious complications occurred in 11 patients (1.3%): vertebral body osteomyelitis (1), epidural abscess with neurologic impairment (2), pulmonary microabscesses from internal jugular vein abscess (1), tracheoesophagogastric anastomotic fistula (1), and major dehiscence necessitating anastomotic takedown (6). These complications became manifest from 5 to 85 days after the esophageal resection and reconstruction (mean 19 days). Leakage from a gastric suspension stitch placed in the anterior spinal ligament over the vertebral bodies resulted in a posterior gastric leak and either osteomyelitis or an epidural abscess in three patients, none of whom had evidence of extravasation on the routine barium swallow 10 days after operation. Cervical exploration for a presumed anastomotic leak led to the unexpected discovery of an abscess formed by the stomach and the adjacent wall of the internal jugular vein, which was ligated and resected. One patient without symptoms who was discharged from the hospital with a contained anastomotic leak on the postoperative barium swallow was readmitted 7 days later with a cervical tracheoesophagogastric anastomotic fistula of which he ultimately died. In 6 patients (7% of those who had anastomotic leaks) there was sufficient gastric ischemia or necrosis, or both, to necessitate takedown of the anastomosis and intrathoracic stomach, cervical esophagostomy, and insertion of a feeding tube. As a result of this experience, it is recommended that cervical gastric suspension sutures either be omitted entirely or placed in the fascia over the longus colli muscles anterior to the spine, but not directly into the prevertebral fascia overlying the vertebral bodies or cervical disks. All but minute cervical anastomotic leaks, even if apparently contained, are best drained rather than treated expectantly. Patients who remain febrile and ill after bedside drainage of a cervical esophagogastric anastomosis leak should undergo cervical reexploration in the operating room; major gastric ischemia or necrosis, or both, may warrant takedown of the anastomosis and intrathoracic stomach. ( J THORAC CARDIOVASC SURG 1995;110:1493-1501)