AIMS: Benign anastomotic stricture (BAS) is a common cause of dysphagia following oesophagectomy. The aim of this study was to assess the incidence of BAS, identify risk factors for its development and evaluate the results of postoperative endoscopic dilatation. METHODS: A consecutive series of 234 patients undergoing oesophagectomy with a stapled intrathoracic oesophagogastric anastomosis (Autosuture CEEA gun) between April 1990 and April 1999 were studied. BAS was defined as dysphagia with anastomotic narrowing (XQ200 endoscope) and no suspicion of recurrence. Statistical analysis was by the chi2 and Mann-Whitney U tests. RESULTS: The postoperative mortality rate was 5 per cent (12 of 234) and the anastomotic leak rate 3 per cent (six of 234). One-third of patients (70 of 222) who survived surgery re-presented with dysphagia; 5 per cent (11 of 222) were found to have proven local recurrence and 27 per cent (59 of 222) BAS. The median time to development of BAS was 92 (range 24-210) days. BAS formation was significantly related to the size of the staple gun employed: 21 mm, 80 per cent (eight of ten); 25 mm, 32 per cent (25 of 78); 28 mm, 23 per cent (19 of 81); 31 mm, 19 per cent (seven of 37) and 34 mm, 0 per cent (none of five) (chi2 = 18.3, 4 d.f., P < 0.01). All patients underwent radiographically controlled endoscopic dilatation with no complications. Recurrent BAS occurred in over half of these patients (32 of 59), who had a median of 2 (range 2-17) recurrences all resolving within 532 (mean interval 68) days. Recurrent BAS formation was also significantly related to the size of the staple gun employed (chi2 = 11.6, 4 d.f., P < 0.05). Following the introduction of the Autosuture 'tilt-top' device in July 1995 the median size of gun used rose from 25 to 28 mm with an overall decrease in the incidence of BAS from 33 to 21 per cent (chi2 = 4.0, 1 d.f., P < 0.05). All patients with anastomotic leaks survived and none subsequently developed BAS. Similarly, no association was found between the development of BAS and the anastomotic site (measured from incisors), tumour subtype, resection margin length, sex, age or preoperative cardiorespiratory status. CONCLUSIONS: Staple gun size is an important risk factor for BAS formation and 'tilt-top' devices enable the use of a larger head with a subsequently lower incidence of BAS. Endoscopic dilatation is an effective treatment for BAS which rarely recurs and always resolves within 18 months.