The political framework and the development of molecular biology and electronic data management caused a general paradigm shift in plant genetic resources (PGR), exemplified here for wheat. (1) In situ versus ex situ maintenance of PGR. Ex situ maintenance lost predominance. Wild wheats are effectively maintained in the wild; landraces do well on farm. New methods did not lead to the expected progress. (2) Inclusion of neglected and underutilized crop species. Some species are probably extinct in traditional cultivation areas, whereas landraces were recently found for others. Wild relatives have gained importance in wheat breeding: besides wild Triticum species, also Aegilops, Secale, Hordeum and other genera are used. ×Triticosecale reached world importance; ×Tritordeum will follow soon. (3) Methods of analysing diversity within and between taxa. New technology yields new insights in the structure and evolution of populations. (4) Genetic erosion is a problem, also inside genebanks. (5) Landraces show complex morphological diversity. Infraspecific classification systems are useful for their characterization and handling, but less recognised by breeders. (6) Methods of evaluation. Molecular markers identify genetic differences on a fairly simple level without reference to ecological adaptation. (7) Genebanks should expand classical evaluation programmes. Pre-breeding will gain importance. (8) Storage and reproduction in genebanks is done effectively and cost-efficiently under long-term conditions, but strategic concepts for reproduction are needed. Traditional methods are often neglected, and modern possibilities over-emphasized. Maintenance of landraces in genebanks and on farm poses challenges. PGR work is conservative. Landraces can be studied by traditional methods; molecular methods can resolve specific questions.