Pancreatic, gastric, and colorectal carcinomas are diagnosed in 200,000 Americans each year. Therapeutic options for patients with advanced disease are limited; conventional chemotherapy is palliative and produces complete responses in only a few patients. Clinical research has focused on the evaluation of investigational new drugs, combination regimens, and biochemical modulation of fluorouracil. Unfortunately, the results of recent phase II studies of new agents have been disappointing. The exception is CPT-11, a topoisomerase I inhibitor, which showed promising activity in colorectal cancer (in including patients who had failed prior fluorouracil therapy). Modified regimens of fluorouracil and methotrexate with either doxorubicin alone or with epirubicin and cisplatin were associated with response rates approaching 50% in patients with gastric cancer, but appeared to be less toxic than previously published regimens. A randomized trial comparing fluororacil alone or with oral leucovorin allowed early dose escalation according to individual tolerance; the response rate to fluorouracil alone (23%) was higher than that reported in previous phase III trials, suggesting the importance of using adequate doses to produce toxicity in terms of clinical response. Hepatic arterial infusion of floxuridine was associated with a 39% response rate in colorectal cancer patients with disease confined to the liver for whom systemic fluorouracil therapy had failed, suggesting this approach is a reasonable therapeutic option in carefully selected patients.