There is clear evidence that genetic information could be used to optimize treatment of cancer patients. The therapeutic index of chemotherapy could be enlarged by reducing the risk of toxicity, increasing the likelihood of tumor response, or both. Although germline DNA information has been already used to identify patients at high risk of toxicity, it is a matter of debate whether it could be also used to predict tumor response. Due to the intrinsic chromosomal and genetic instability of the tumor genome, recent evidence supports the concept that tailoring of chemotherapy in cancer patients might be better achieved by mutational analysis of patient tumor DNA. However, a proportion of germline variation is still retained in the DNA of cancer cells. Hence, we cannot exclude that germline genetic variation might be informative of a particular tumor phenotype. This review will provide a thorough analysis of the role of germline and tumor DNA variation in cancer pharmacogenomics. It will discuss the prediction of toxicity risk by means of germline genetic make up of patients, using irinotecan- UGT1A1 and 6-mercaptopurine-TPMT as two examples. This review will also provide evidence of the role of tumor mutational analysis to predict tumor response and the emergence of clinical resistance to therapy, using the examples of 1) EGFR somatic mutations in lung cancer patients treated with EGFR-inhibitors, and 2) BCR-ABL and KIT somatic mutations in chronic myeloid leukemia and gastrointestinal stromal tumor patients treated with imatinib. The clinical relevance of loss of heterozygosity in tumor samples will be elucidated by describing the findings on the thymidylate synthase gene in colorectal cancer patients treated with fluoropyrimidines. The karyotypic abnormalities in the TPMT in acute lymphoblastic leukemia patients provide evidence that the germline genotype of cancer cells might be different from the cellular phenotypes. In addition, the clinical impact of the germline TPMT and CYP2D6 genetic variation for patient response will be discussed. Finally, the role of gene amplification and its interplay with somatic mutations of the EGFR in lung cancer patients will be reviewed.