The patient (germline) genome can influence the pharmacokinetics and pharmacodynamics of cancer therapy. The field of pharmacogenetics (PGx) has primarily focused on genetic predictors of pharmacokinetics, largely ignoring pharmacodynamics, using a candidate approach to assess single-nucleotide polymorphisms (SNP) with known relevance to drug pharmacokinetics such as enzymes and transporters. A more comprehensive approach, the genome-wide association study, circumvents candidate selection but suffers because of the necessity for substantial statistical correction. Pharmacogene panels, which interrogate hundreds to thousands of SNPs in genes with known relevance to drug pharmacokinetics or pharmacodynamics, represent an attractive compromise between these approaches. Panels with defined or customizable SNP lists have been used to discover SNPs that predict pharmacokinetics or pharmacodynamics of cancer drugs, most of which await successful replication. PGx discovery, particularly for SNPs that influence drug pharmacodynamics, is limited by weaknesses in both genetic and phenotypic data. Selection of candidate SNPs for inclusion on pharmacogene panels is difficult because of limited understanding of biology and pharmacology. Phenotypes used in analyses have primarily been complex toxicities that are known to be multifactorial. A more measured approach, in which sensitive phenotypes are used in place of complex clinical outcomes, will improve the success rate of pharmacodynamics SNP discovery and ultimately enable identification of pharmacodynamics SNPs with meaningful effects on treatment outcomes.