Thirty-seven previously untreated patients with advanced metastatic colorectal carcinoma were treated in a prospective randomized fashion with either ICRF-159 or 5-FU. The ICRF-159 was administered orally at a dose of 1 g/m2/day for 3 consecutive days every 3 weeks, and the 5-FU was given iv at a dose of 450 mg/m2/day for 5 days every 5 weeks. All patients were evaluated for response and toxic effects after two courses of treatment. All those who failed to meet the criteria for objective response with either a complete remission or a partial response received the other drug in a crossover fashion. Three of the 18 patients (16%) initially treated with 5-FU achieved a partial response while none of the 19 patients initially treated with ICRF-159 achieved a complete or partial response. Nine prior 5-FU-treated patients were crossed over to ICRF-159 and 14 prior ICRF-159-treated patients subsequently received 5-FU. No antitumor response was seen with the secondary agent in this study. The response rate for ICRF-159 (none of 19 patients) predicts that it is unlikely to produce a true response rate of greater than or equal to 20% with a rejection error of less than 5%, making it unsuitable as primary therapy for colon carcinoma. The toxicity of 5-FU was moderate and mainly gastrointestinal while the toxicity of ICRF-159 was severe and mainly hematologic.