A prospective randomized trial of 67 patients undergoing oesophagectomy by either a transhiatal approach or right-sided thoracotomy was conducted over a 40-month period. In 32 patients the approach was transhiatal and 35 had a thoracotomy; the groups were well matched before operation. There were two hospital deaths in patients having the transhiatal oesophagectomy and three in those undergoing thoracotomy. Blood transfusion, intensive care stay and overall time of hospitalization were not significantly different between the two groups. There was no difference in the postoperative morbidity rate and, in particular, the incidences of pulmonary complications were similar (19 per cent for transhiatal oesophagectomy, 20 per cent for thoracotomy) with anastomotic fistula in 6 and 9 per cent respectively. The median (range) operating time was significantly longer in patients having thoracotomy (6 (3.5–9.5) versus 4 (3–8) h). Long-term survival was unaffected by the type of operation performed or addition of preoperative chemotherapy or radiotherapy. Nodal status was a signficant prognostic factor within but not between the two groups. It is concluded that oesophagectomy by a transhiatal route or right thoracotomy are equally effective surgical options for treatment of squamous cell oesophageal cancer.