Breeding endangered mammals for their conservation requires knowledge about the genetic architecture of the respective species. In taxa with tight genetic cohesion between populations, the definition of management units for captive breeding rarely poses problems, except if there are morphologically well differentiated subspecies grading into one another although they are hardly separate at the molecular level. Species with genetic diversity predominantly between populations can pose serious problems for breeders. Examples are discussed of mammalian species with complex genetic architectures, where decisions have to be drawn whether to select only certain populations for conservation, or to create an artificial taxon. Research into subspecific molecular taxonomy of rare zoo-living wildlife is frequently hampered by small sample sizes available for study, with the risk of spurious molecular taxonomie distances based on marker allele frequencies in populations influenced by genetic drift. "Typological" approaches are suggested for molecular systematics of such study material, with the haplotype organization of polymorphic MHC genes appearing particularly promising. Additional molecular approaches, not easily susceptible to sample size problems, are shortly presented. The implementation of breeding plans to achieve conservation genetic goals may interfere with the social structures of the animals. This group of problems includes the transfer of socially compatible individuals to form new groups, and the provision of suitable sexual partners for mate choice mechanisms to act in their species-specific manner. A survey of scientific research in European zoos is provided to recognize what is being done to fight the current ignorance about basic aspects of the conservation biology of many endangered mammals.