In a retrospective study of 127 patients who underwent resection for oesophageal cancer without anticoagulant thromboprophylaxis, the incidence of clinically established postoperative deep vein thrombosis (DVT) was 1.6 per cent. In a prospective study the Sue-Ling DVT risk factor index was calculated for 53 consecutive patients with oesophageal cancer; the mean (s.d.) value was -4.4(3.5). For 29 patients who underwent resection the following policy of thromboprophylaxis was applied. Patients undergoing transthoracic oesophagectomy with risk score > -4 (four patients) and those undergoing transhiatal resection with risk score > -1 (four) received both anticoagulant and mechanical thromboprophylaxis; the remaining 21 patients received only mechanical thromboprophylaxis. No patient developed DVT after operation. There were no haemorrhagic complications in the group undergoing anticoagulant thromboprophylaxis. The selective approach, based on the Sue-Ling risk factor index and type of operation, spared 72 per cent of the patients operated on from undergoing anticoagulant thromboprophylaxis and produced favourable results in this small series.