The results of two schedules of the combination of etoposide, doxorubicin, and cisplatin (EAP) in gastric cancer are reported. EAP-1 was administered as originally reported and consisted of i.v. doxorubicin (adriamycin) 20 mg/m2 on days 1 and 7, i.v. cisplatin 40 mg/m2 on days 2 and 8 and i.v. etoposide (VP-16) 120 mg/m2 (100 mg/m2 in patients aged 60-65) on days 4-6. EAP-2 consisted of i.v. adriamycin 40 mg/m2 on day 1, i.v. cisplatin 80 mg/m2 (total dose per cycle) given in 3 divided doses on days 1-3 and i.v. VP-16 100 mg/m2 on days 1-3. Cycles of the two regimens were repeated on day 22. Drug doses were reduced in patients over 65 years of age. Twenty patients were treated with EAP-1 and 43 with EAP-2. Forty-five of the 63 patients included in this study had advanced gastric carcinoma, 16 had radically resected stage III disease, and 2 had metastatic signet ring cell carcinoma of unknown primary origin. Thirty-eight patients with advanced gastric adenocarcinoma were evaluable for response. The response rate for EAP-1 (6/13, 46%) was similar to that of EAP-2 (13/25, 52%). The median duration of response was 8 months for EAP-1 and 6.5 months for EAP-2. Myelotoxicity of EAP-1 was much more severe than that of EAP-2. Hospitalization due to granulocytopenic fever was required in 15/78 (19%) EAP-1 versus 20/215 (9%) EAP-2 courses. Toxic deaths occurred in 3/20 treated with EAP-1 and in 1/45 treated with EAP-2. The difference in toxicity between the two regimens could not be attributed to differences in patients’ characteristics or to dose intensity. We conclude that the modified EAP regimen (EAP-2) is effective in the treatment of gastric cancer and is less toxic than the original EA