Phase II studies of fluorouracil (5-FU) administered by protracted intravenous infusion have suggested an improved response rate and decreased toxicity profile when compared with 5-FU given by bolus injection in patients with metastatic colorectal cancer. Additional studies have suggested further enhancement of infusion 5-FU activity when it is combined with low-dose weekly cisplatin administration. This phase III study in adults with metastatic colorectal cancer was planned as a comparison of objective response rates, toxicity, and survival in patients receiving bolus versus protracted-infusion 5-FU with or without cisplatin. Four hundred ninety-seven previously untreated patients with advanced, measurable metastatic colorectal cancer were randomly assigned to receive treatment A (bolus 5-FU at 500 mg/m 2 for 5 days followed in 2 weeks by weekly bolus 5-FU at 600 mg/m 2 ), treatment B (bolus 5-FU at 500 mg/m 2 for 5 days followed in 2 weeks by weekly bolus 5-FU at 600 mg/m 2 , plus weekly cisplatin at 20 mg/m 2 ), treatment C (5-FU at 300 mg/m 2 per day by continuous infusion), or treatment D (5-FU at 300 mg/m 2 per day by continuous infusion plus weekly cisplatin at 20 mg/m 2 ). All drugs were administered intravenously. Enrollment in the trial occurred from August 1987 through December 1990, and follow-up was through September 1995. The Kaplan-Meier method was used to estimate overall and disease-free survival, and Cox regression models were used to assess the effects of patient characteristics on survival. All P values resulted from two-sided tests. Objective tumor response was observed in 28 (18%) of 153 patients receiving treatment A, in 45 (28%) of 159 patients receiving treatment C (C versus A; P = .045), and in 47 (31%) of 153 patients receiving treatment D (D versus A; P = .016). Because of excessive toxicity, treatment B was discontinued after only 12 patients had begun treatment. Median time to disease progression was 5.1 months for patients in arm A compared with 6.2 and 6.5 months for patients in arms C and D, respectively (C versus A, P = .007; D versus A, P = .017). Patterns of toxic effects differed substantially among the treatment arms. Forty-five percent of the patients receiving bolus 5-FU alone (A) experienced grade 3–4 leukopenia, with two sepsis-related deaths. Hand-foot syndrome and mucositis were the major treatment-limiting toxic effects for patients in the two treatment arms involving infusion. Despite the improvement in response rates and time to disease progression with infusion 5-FU with or without cisplatin (C and D, respectively) ( P = .003), the overall survival for the three groups (A, C, and D) was similar ( P = .307). This may have been due in part to a longer median survival time of 10.4 months for patients in arm A, compared with an anticipated survival of 7 months. 5-FU given as a continuous infusion produced a higher objective response rate, a modest prolongation in time to disease progression, and less life-threatening myelosuppression in patients than bolus 5-FU. Concomitant treatment with low-dose cisplatin caused added toxicity and complexity of treatment and did not provide a major clinical benefit. No statistically significant survival differences were observed among the three treatment groups. [J Natl Cancer Inst 1996;88:668–74]