Dose adjustment of drug administration for each patient has been performed based on counts of some factors such as body surface area, age of the patient, performance status, renal and/or liver function. Pharmacokinetic and pharmacodynamic analyses have been investigated by measuring the plasma concentration of a drug and observing the drug effects. However, prior to drug administration it is difficult to predict unexpected, severe drug toxicity, which depends on the individual differences among patients. Recent progress in human genome analysis has been providing tools for new approaches to disease treatment based on individual differences using genetic information. This review focuses on the drug metabolizing enzyme and its genetic polymorphisms in cancer chemotherapy. We describe the recent findings on pharmacogenomics between toxicity and the genetic polymorphisms of the thiopurine methyltransferase (TPMT) gene, dihydropyrimidine dehydrogenase (DPYD) gene, methylenetetrahydrofolate reductase (MTHFR) gene, and uridine diphosphate glucuronosyltransferase (UGT1A1 and UGT1A7) gene. We need to accumulate clinical data based on the variation of genetic profiling as well as pharmacogenetic information. Such data will help tailor cancer chemotherapy to an individual's predisposition in the near future.