Field experiments to test the biomass production from the C-4 perennial grass Miscanthus have concentrated on one triploid genotype, namely M. x giganteus. Several limitations to production from M. x giganteus including insufficient winter rhizome cold tolerance demonstrated the need to broaden the genetic base. This paper presents results from a field trial in Southern Germany planted with 15 Miscanthus genotypes including M. x giganteus, M. sacchariflorus, wild M. sinensis and bred M. sinensis hybrids over 3 years. Under field conditions, establishment from micro-propagation was high and all genotypes survived the first winter (losses <6%). Genotype growth characteristics were determined by measurements of height, shoot density, stem diameter. flowering time and autumn senescence rate. Increasing plant age was associated with higher yields (2, 6, and 17 t dry matter ha(-1) in autumn of years one to three, respectively) and better biomass qualities for combustion. Delaying harvest time from autumn until spring in the second and third years reduced harvestable yields but increased combustion quality by lowering moisture, ash, Cl and N contents for all genotypes. Yields were highest for the M. x giganteus and some newly bred M. sinensis hybrids, but biomass qualities were best in the pure M. sinensis genotypes. These trials showed that new hybrids between M. sacchariflorus and M. sinensis should be developed that have many growth characteristics similar to M. x giganteus but with improved rhizome freeze tolerance in winter. As the biomass yield and quality of a particular genotype after I year of growth does not always relate to that measured in subsequent years, identification of the most suitable genotypes requires at least 2 years. (C) 2002 Elsevier Science B.V. All rights reserved.