Between 1990 and 2004, 149 patients underwent esophagectomy with thoracotomy at our institution. Because the retention of a nasogastric tube postoperatively impedes expectoration, this practice was abolished in 1997, in order to reduce the risk of respiratory tract infection (RTI). Since 1998, we have instead performed gastrostomy for decompression. In this retrospective study, we compared the incidence of postoperative respiratory tract infection between these two groups. We previously demonstrated that the concordance between bacteria detected in the gastric juices and those detected in sputum was more than 70% in patients with postoperative RTI. In the present study, the incidence of postoperative respiratory tract infection was significantly higher in patients in whom a nasogastric tube was retained when compared to the rate in those for whom gastrostomy was performed instead (41.5% [34/82] vs 26.9% [18/67]; P = 0.045). The rate of other infections did not differ significantly between the two groups. We conclude that, in the postoperative management of esophagectomy, the retention of a nasogastric tube impedes expectoration. In addition, nasogastric tubes appear to increase the risk of postoperative RTI, considering the previous finding that the concordance of bacteria in sputum and gastric secretion was over 70%.