Fluorouracil is used clinically against various solid tumours. Both fluorouracil toxicity and pharmacokinetics vary highly within and between individuals. The reasons why doses are not individualised routinely are difficulties in defining, predicting and achieving an optimal fluorouracil exposure or dose because of a narrow therapeutic index, nonlinear pharmacokinetics, variabilities in administration rates and metabolism, and in targets like thymidylate synthase. To individualise fluorouracil administration before the first dose, assessment of the individual dihydropyrimidine dehydrogenase (DPD) activity may be useful, because this genetically highly polymorphic enzyme controls approximately 80% of fluorouracil elimination. A complete or partial loss of DPD activity in 0.1 and 3-5% of Caucasians, respectively, leads to increased fluorouracil exposure and toxicity. Several methods to assess DPD activity in patients have been proposed (genotyping, various phenotyping methods), but each of them has limitations, as has the fluorouracil test dose approach. To adapt exposure towards fluorouracil a priori, a combination of genotyping and phenotyping may yield better prediction of toxicity than one method alone. A prerequisite for dose adaptation is the definition of fluorouracil exposure ranges with sufficient therapeutic activity, but without serious toxicity. While an increased risk of leukopenia, diarrhoea, stomatitis, and hand-foot syndrome during continuous 5-day infusions was related to fluorouracil exposures above an area under the plasma concentration-time curve (AUC) threshold of 25-30 mg.h/L, tumour response was higher when an AUC of approximately 30 mg.h/L was achieved, illustrating the extremely narrow therapeutic window of fluorouracil. Pharmacokinetic target values are less clear for other regimens, including chronomodulated regimens, which yielded a superior clinically efficacy and tolerability in several trials. However, the monitoring of fluorouracil plasma concentrations seems principally useful for individual a posteriori dose adjustment. Whether an adaptation of the fluorouracil starting dose to the results of two DPD activity tests before fluorouracil administration a priori, and the adaptation of doses to fluorouracil exposure a posteriori is a reasonable approach to better prevent toxicity and increase efficacy, remains to be evaluated in randomised clinical studies comparing these strategies to routine clinical safety monitoring.