The carbide fuel which will be introduced into the SNR for the first time after 1980, will be helium bonded, with a low linear heat rating and fuel density. This design appears to be the least problematic one for a medium burnup goal. The restriction to a moderate design arises to a certain extent from economical and safety reasons, but mainly can be attributed to uncertainties in extrapolation to a very high performance level. This can be demonstrated in a general discussion valid for both oxide and carbide. Due to the fact that the fuel elements of a large power plant with a peak burnup of 100 MWd/kg will undergo radiation damages, which cannot be demonstrated by experimental fuel pins in test (DFR, Rapsodie) or demonstration reactors (SNR, PFR, etc.) for the first generation, licensable maximum burnups will be in the range of 70 MWd/kg. It is impossible to perform tests outside the future large power plants with a relevant neutron dose-to-fuel burnup ratio. Therefore, in the German program a continuous development has been underway since 1968, in which separate medium burnups and neutron doses are demonstrated in test and demonstration reactors. In addition, the fuel fabrication process will be steadily improved in a pilot fabrication plant.