Pharmacogenomics is the study of how variation in genes can lead to variation in the response to drugs. In the context of the complete human genome sequence, we can begin to formulate strategies for personalized medicine, based on the detailed sequence of the genes for individual patients. Pharmacogenomics is thus one of the major ways in which personalized medicine may become a reality. The current challenge to pharmacogenomics research is to gather information about the location and frequency of variations in human genome sequences, and information about the associated laboratory and clinical phenotypes manifested by these individuals, and to understand how the two correlate. In the future, we will genotype patients before prescribing medicines in order to increase efficacy and decrease side effects. As part of this international effort in pharmacogenomics, we are building the Pharmacogenomics Knowledge Base (PharmGKB) which serves the research community as a resource for genotype and phenotype data, and visualization and analytic tools to analyze them. In this talk, I will describe the current PharmGKB, as well as the challenges facing it in literature analysis and curation, ontology definition, and provision of analytic functionalities.