The use of fluorouracil has been complicated by unpredictable pharmacokinetics, low response rates and seemingly random toxicity. The variable pharmacology is largely due to inherited differences in expression of the metabolising enzyme dihydropyrimidine dehydrogenase (DPD). This converts fluorouracil to inactive metabolites (catabolic pathway) and ultimately dictates the amount of fluorouracil that is available to be metabolised to cytotoxic nucleotides (anabolic pathway). Absolute and partial DPD deficiency affect around 0.1 and 3% of the Caucasian population, respectively. Administration of conventional doses of fluorouracil to these individuals has resulted in profound bone marrow and gastrointestinal toxicity, especially in those with absolute DPD deficiency. Other forms of toxicity such as myocardial ischaemia have been difficult to attribute directly to DPD deficiency. Efforts to improve outcomes with fluorouracil have included monitoring of fluorouracil concentrations and modifying fluorouracil administration techniques (e.g. from bolus injections to protracted intravenous infusions). In general, these moves have met with limited therapeutic advancement. The recognition that DPD deficiency increases toxicity has lead to the suggestion that genotypic or phenotypic assessment of DPD status prior to initiating fluorouracil may help predict outcomes. The gene that encodes for DPD expression is called DPYD. Approximately 1% of Caucasians are heterozygotes for the DPYD*2A mutation which is the variant allele that is most frequently implicated in DPD deficiency. Screening for this mutation may identify around 60% of individuals with absolute DPD deficiency who are at the greatest risk of toxicity. Another approach is to determine DPD activity in peripheral blood mononuclear cells, with low activity suggesting an increased risk of toxicity. Intratumoral DPD activity may also be assessed with high activity suggesting a poorer response to fluorouracil. Recently, drugs that inhibit DPD (e.g. eniluracil) have become available. These remove much of the variability in fluorouracil pharmacokinetics and may make assessment of DPD activity redundant. Despite the considerable inroads that have been made, further study is needed before the best means of optimising fluorouracil treatment is determined.