The purpose of this phase II randomised trial was to determine which of two schemes, raltitrexed-irinotecan or raltitrexed-oxaliplatin, offered better activity and less toxicity in patients with advanced colorectal cancer (CRC). A total of 94 patients with previously untreated metastatic CRC were included and randomised to receive raltitrexed 3 mg m(-2) followed by oxaliplatin 130 mg m(-2) on day 1 (arm A), or CPT-11 350 mg m(-2) followed by raltitrexed 3 mg m(-2) (arm B). In both arms treatment was repeated every 3 weeks. Intent-to-treat (ITT) analysis showed an overall response rate of 46% (95% CI, 29.5-57.7%) for arm A, and 34% (95% CI, 19.8-48.4%) for arm B. Median time to progression was 8.2 months for arm A and 8.8 months for arm B. After a median follow-up of 14 months, 69% of patients included in arm A were still alive, compared to 59% of those included in arm B. Overall, 31 patients (65%) experienced some episode of toxicity in arm A and 32 patients (70%) in arm B, usually grade 1-2. The most common toxicity was hepatic, with 29 patients (60%) in arm A and 24 patients (62%) in arm B, and was grade 3-4 in four (8%) and four (9%) patients, respectively. In all, 14 patients (29%) from arm A and 24 patients (52%) from arm B had some grade of diarrhoea (P<0.03). Neurologic toxicity was observed in 31 patients (64%) in arm A, and was grade 3-4 in five patients (10%), while a cholinergic syndrome was detected in nine patients (19%) in arm B. There were no differences in haematologic toxicity. One toxic death (2%) occurred in arm A and three (6.5%) in arm B. In conclusion, both schemes have high efficacy as first-line treatment in metastatic CRC and their total toxicity levels are similar. Regimens with raltitrexed seem a reasonable alternative to fluoropyrimidines.