The purpose of this study was to evaluate the safety and activity of fixed-dose capecitabine in patients with advanced colorectal cancer and to correlate pretreatment plasma concentrations of homocysteine and serum and red cell folate with toxicity. Patients received capecitabine 2000 mg (4 x 500 mg tablets) twice daily on days 1-14 every 3 weeks. They were reviewed weekly during the first cycle and then three weekly for safety assessment. Eligibility criteria were advanced/metastatic colorectal cancer, < or = 2 prior chemotherapy regimens, ECOG performance status 0-2 and life expectancy >12 weeks. A total of 60 patients were enrolled and 55 were evaluable for efficacy. The median age was 72 years and 63% of patients had a performance status of 1 or 2. Confirmed tumour responses were reported in 15 patients (28%; 95% confidence interval (CI), 15.7-40.3%). The median time to disease progression was 4.9 months and median overall survival was 11.2 months. The median ratio of fixed dose to body surface area (BSA)-calculated dose was 88% (range 65-108%). Significant myelosuppression was not observed. Grade 2/3 treatment-related adverse events were diarrhoea (34%), fatigue (27%), stomatitis (15%) and hand-foot syndrome (22%). Dose reduction due to adverse events was required in 16 patients (29%) and multiple reductions in five patients (9%). There was no grade 3/4 haematological toxicity, any grade 4 adverse events or treatment-related deaths. Patients with higher pretreatment levels of serum folate experienced significantly greater toxicity (P = 0.02, CI: 1.0-1.2) during cycle 1 and over the entire treatment period (P = 0.04, CI: 1.0-1.3). Pretreatment homocysteine concentrations did not predict for toxicity. In conclusion, fixed-dose capecitabine appears to have similar efficacy and safety compared to the currently recommended dose schedule based on body surface area and simplifies drug administration. A high pretreatment folate may be predictive of increased toxicity from capecitabine.