PVI 5FU gives increased response rates and reduced toxicity when compared to bolus 5FU (J Clin Oncol 1989, 425–432). PVI 5FU administration was reported to give highly variable (>1000-fold) plasma 5FU concentrations at steady state (FU Css) which correlated with toxicity (Ann Oncol 1996, 47–53); but only 19 patients were studied. Therefore, we performed a study of PVI 5FU in 61 patients with advanced colorectal cancer to assess the variability (inter- and intra-subject) in 5FU Css associated with PVI 5FU (300 mg m−2 day−1) and to attempt to correlate pharmacodynamic end-points (anti-tumour activity, toxicity) with 5FU Css as a prelude to ‘exposure-guided’ 5FU administration. All 5FU sampling was performed between 10 am and noon. PVI 5FU administration continued to 26 weeks in patients with disease improvement or stabilization. The response rate was 26% (33% stable disease) and median survival was 11 months. Hand–foot syndrome was the most common dose limiting toxicity. Variability in 5FU300Css was considerably less than previously reported; 94 ± 25 ng ml−1(CV = 27%). No relationships were demonstrated between subject mean 5FU300Css and PD end-points such as response, mucositis, diarrhoea and hand–foot syndrome. The lack of correlation suggests that measurement of 5FU concentrations should not be used to individualize dosing in patients receiving PVI 5FU for advanced colorectal cancer.