In a prospectively randomized trial, patients with advanced locally recurrent or metastatic gastric adenocarcinoma were randomized to receive 5-fluorouracil (5-FU) and methyl-CCNU; 5-FU, Adriamycin (Adria Laboratories, Columbus, Ohio), and methyl-CCNU; 5-FU, Adriamycin, and mitomycin C; or Adriamycin and mitomycin C alone. One hundred eighty-three previously untreated evaluable patients were randomized among the four arms. An additional 39 patients previously treated with 5-FU, were assigned to treatment directly to Adriamycin and mitomycin C. Response rates were 14%, 29%, 39%, and 29%, respectively, among previously untreated patients and 21% for Adriamycin and mitomycin C among previously treated patients. 5-Fluorouracil, Adriamycin, and mitomycin C, the arm containing the largest number of responders (18), was the combination associated with the longest median survival. A larger proportion of patients in this arm survived one year or more. In addition, the 5-FU, Adriamycin, and mitomycin C program had the lowest rate of severe or worse toxicity of any of the treatments and was effective in patients who were less than fully ambulatory and in those who had lost weight. 5-Fluorouracil, Adriamycin, and mitomycin C appear to be a likely combination to be considered in a surgical adjuvant program.