Cancer chemotherapy is characterized by a broad range of efficacy and toxicity among patients. Most anticancer drugs show wide interindividual variability in pharmacokinetics and have narrow therapeutic windows. Since drug metabolism is often an essential determinant of interindividual variability in pharmacokinetics, pharmacogenomic studies of drug-metabolizing enzymes are expected to rationalize cancer chemotherapy in terms of patient, treatment, and dosage selection. Candidate gene approaches to pharmacogenomics are based on existing knowledge in clinical pharmacology, used to select the target(s) to be analyzed. So far, the candidate gene approach has provided important clues for pharmacogenomic-based personalized chemotherapy with 6-mercaptopurine (6-MP), solely metabolized by thiopurine S-methyltransferase (TPMT), and irinotecan, mainly detoxified by UDP-glucuronosyltransferase 1A1 (UGT1A1). Reduced activity of TPMT caused by polymorphisms in the TPMT gene and decreased activity of UGT1A1 caused by UGT1A1*28 are related to severe toxic effects of 6-MP and irinotecan, respectively. In response to these findings, the Food and Drug Administration in the United States has supported clinical pharmacogenetic testing by revising the package inserts for these anticancer drugs. The genome wide approach to pharmacogenomics has gradually evolved with continued progress in genome sciences and technologies. This approach can disclose previously unknown relations of factors, as well as identify potential multigenetic associations. The genome wide approach can also identify genes underlying the phenotypic effects of anticancer drugs. This approach may play a complemental role to the candidate gene approach in the future of cancer pharmacogenomics. This review describes recent progress in pharmacogenomics in the field of cancer chemotherapy.