The approximate parameters of steady-state D-T fusion reactors can be predicted on the basis of present knowledge, provided some interim assumptions are made. The most speculative assumption is plasma confinement: either about 100 Bohm times in a closed magnetic system whose minor plasma radius is several metres; or about 10 effective 90° Coulomb scattering times in an open system of similar size. Almost as serious is the assumption that a flux 3 × 1018 m−2 sec−1 of 14-MeV neutrons will not cause excessive radiation damage to the device. A simple plasma model includes particle injection with or without energy, particles loss, electron bremsstrahlung and synchrotron radiation, thermalization of alpha particles (from fusion) and energy exchange between all species. It predicts equilibrium temperatures T, pressures, fractional burn-up per confinement time fb and other quantities. For open-ended systems fb approaching 0.1 is required, which is several times that obtainable against simple Coulomb scattering; some possibilities for ameliorating this difficulty remain unexplored. Typically Tb ≈ 60 to 80 keV, but Te ≈ 30 to 40 keV because of radiation. For closed systems fb can be lower, but fb ≈ 10% with artificially enhanced bremsstrahlung would be more desirable. In every case, alphas are thermalized in the plasma. Using niobium structural material and mostly lithium moderator, one can obtain a (regenerated triton/incident neutron) breeding ratio between 1.2 and 1.6, depending upon the configuration. Cost per unit power would be very low for large systems. One intermediate example gives $6/kW(th); β = 0.15 (closed system), 0.34 (open system), for a device with 3-m radius and 10 T maximum field. Total power could exceed 10 000 MW(th). The large neutron excess in such devices would be very valuable and there are other advantages.