We conducted a randomized trial to compare the safety and effectiveness of aprepitant, granisetron, and dexamethasone (AGD) with those of palonosetron and dexamethasone (PD) in patients who received highly emetogenic chemotherapy (HEC). Patients with esophageal or gastric cancer who were scheduled to receive HEC including at least 60 mg/m2 of cisplatin as the first-line treatment were randomly assigned to receive AGD (oral aprepitant 125 mg on day 1 and 80 mg on days 2–3; intravenous granisetron 3 mg on day 1; intravenous dexamethasone 6.6 mg on day 1 and oral dexamethasone 4 mg on days 2–3) or PD (intravenous palonosetron 0.75 mg on day 1; intravenous dexamethasone 13.2 mg on day 1 and oral dexamethasone 8 mg on days 2–3). The primary endpoint was a complete response during the overall study period (0–120 h after the start of chemotherapy) in the first cycle. Eighty-five patients were enrolled, and 84 were eligible. The complete response rate did not differ between the treatment groups, but the proportion of patients with no vomiting was significantly higher in the AGD group than in the PD group (81.4 vs. 58.5%; P=0.031). The results of a quality-of-life survey indicated that the proportion of patients with no or minimal impact on daily life in the vomiting domain was significantly higher in the AGD group (79.1 vs. 53.7%; P=0.020). The primary endpoint of complete response was not achieved, but AGD seems to be more effective than PD for the prevention of HEC-induced vomiting.