Objectives Despite increasing trends toward the early initiation of oral feeding after gastrointestinal (GI) surgeries, current evidence has not been convincing. The present randomized clinical trial aimed to compare the clinical outcomes of early oral feeding (EOF) with late oral feeding (LOF) following surgery for upper GI tumors. Methods One hundred and nine consecutive patients with esophageal or gastric tumors undergoing surgical resection in two hospitals in Tehran, Iran, were enrolled in this prospective randomized controlled trial, and were randomly assigned to a group starting EOF on the first postoperative day and another group that remained nil by mouth until the return of bowel sounds (LOF group). The clinical and surgical outcomes were compared between the two groups. Results The clinical outcomes were significantly better in the patients in the EOF group ( p  < 0.05). Repeated nil per os (14.8 vs. 30.9 %) and re-hospitalization (1.8 vs. 7.3 %) were more common in LOF group ( p  < 0.0001). Additionally, gas passage, nasogastric tube (NGT) discharge, a decrease in intravenous serum to less than 1000 ml per day, the time to start a soft diet and hospital discharge following surgery occurred significantly earlier in the EOF group than in the LOF group ( p  < 0.0001). Conclusion Early oral feeding after the surgical resection of esophageal and gastric tumors is safe, and is associated with favorable early in-hospital outcomes and a sooner return to physiological GI function and hospital discharge.