Abstract. -Darwin Core (DwC) defines a standard set of terms to describe the primary biodiversity data. Primary biodiversity data are data records derived from direct observation of species occurrences in nature or describing specimens in biological collections. The Darwin Core terms can be seen as an extension to the standard Dublin Core metadata terms. The new Darwin Core extension for genebanks declares the additional terms required for describing genebank data sets, and is based on established standards from the plant genetic resources community. The Global Biodiversity Information Facility (GBIF) provides an information infrastructure for biodiversity data including a suite of software tools for data publishing, distributed data access, and the capture of biodiversity data. The Darwin Core extension for genebanks is a key component that provides access for the genebanks and the plant genetic resources community to the GBIF informatics infrastructure including the new toolkits for data exchange. This paper provides one of the first examples and guidelines for how to create extensions to the Darwin Core standard.Keywords. -Darwin Core; Darwin Core extension; GBIF; genebank collections; germplasm; plant genetic resources.There are more than 1750 genebanks distributed all around the world, with more than 130 large and medium-size genebank collections holding more than 10,000 accessions each (FAO, 2010). Each of these genebanks maintains living material of plant genetic resources. New accessions are added to the genebank collections from collecting expeditions and from old cultivars obsolete to the commercial seed trade. The genebank documentation systems are continuously being extended with new accessions and with updated information on existing accessions.The International Treaty on Plant Genetic Resources for Food and Agriculture 1 (ITPGRFA; FAO, 2009, page 29, Article 17.1) calls for building a global information system on plant genetic resources for food and agriculture. Such a system needs to be frequently refreshed with new and updated information from each genebank collection (and other information sources such as inventories of crop wild relatives). With modern information technology, a distributed information system can be designed to allow extracting a snapshot of the decentralized genebank data sets at any time. Moreover, a distributed germplasm information system will make updated germplasm information * Corresponding author; email: dag.endresen@gmail.com. 1 http://www.planttreaty.org/.more easily accessible to plant breeders, crop scientists and other users, thus providing better access also to the plant material. The lack of easy access to germplasm information remains an important bottleneck for utilization of plant genetic resources material . The Global Biodiversity Information Facility (GBIF) 2 has developed a software toolkit for publishing decentralized biodiversity data sets, known as the "Integrated Publishing Toolkit" (GBIF IPT) 3 . It provides a successful example of a software tool ...