BACKGROUND Because of experimental and preliminary clinical evidence that additional modulation of the biochemical pharmacology and cytotoxicity of 5-fluorouracil (5-FU) and leucovorin (LV) may be possible by combination of these agents with cisplatin (CDDP), the authors undertook a prospective randomized trial in patients with colorectal cancer. METHODS Between 1989 and 1992, 138 patients with advanced measurable colorectal cancer previously unexposed to chemotherapy were randomly assigned to treatment with either 5-FU (425 mg/m2) and LV (20 mg/m2) for 5 days, or the combination of 5-FU and LV in the same daily dose plus cisplatin (20 mg/m2), each drug given for 4 consecutive days. In both treatment arms, courses were administered every 28 days, if toxicity allowed, for a total of 6 months or until evidence of tumor progression. RESULTS The overall responses (complete and partial response) were 19% and 28% for the 5-FU/LV and the 5-FU/LV/CDDP treatment arms, respectively. Although the three-drug combination appeared superior to 5-FU/LV for time to progression or death (8.5 versus 5.2 months; P = 0.042), there was no evidence that the adoption of cisplatin will translate into a definite survival advantage. A comparative analysis of the toxicities experienced by the patients in the two treatment groups showed a comparable rate, although severe side effects (P < 0.05), specifically stomatitis (P = 0.013), were noticed more frequently in patients treated with 5-FU/LV for 5 days. CONCLUSIONS These results suggest that the therapeutic index of 5-FU/LV in metastatic colorectal cancer may be improved with the addition of cisplatin. However, the somewhat better therapeutic activity and lower incidence of severe gastrointestinal side effects have to be weighed against additional pharmaceutical charges and the need for a more intense antiemetic regimen.