T earliest serious study of the possibility of utilizing nonequilibrium ionization in MHD generators was carried out by Karlowitz and Halasz at the Westinghouse Research Laboratories between 1938 and 1947. It was originally oriented to the development of a generator that could use combustion products at moderate temperatures. The study was terminated when it was realized that recombination rates in molecular gases were too high to permit useful electron densities to be developed. It was concluded, however, that nonequilibrium ionization might be feasible in the atomic working gases that can be used with a nuclear reactor. This work was reviewed by Karlowitz in 1962. With recent developments in nuclear reactor technology, there is now renewed interest in nuclear-powered MHD generators. The temperature range, which is likely to be accessible with reactors and also desirable thermodynamically, is from about 1500° to 2000° K. It seems likely that cycles with over-all efficiencies between 0.50 and 0.60 can be evolved with such maximum temperatures. The requirements of the reactor and the generator together dictate that the pressure in the generator be between 1 and perhaps 10 atm, if the working fluid is argon and the magnetic field is limited to, say, 20,000 gauss. With the development of large superconducting magnets, it may be possible to raise these pressures by a factor of 5. It is likely that the inert gas will be seeded with a fraction of a percent of cesium. Since 1960, considerable effort has been devoted to the study of such plasmas. It began with small-scale studies of the behavior of the plasma, and has only recently advanced to the study of complete generators, the point at which Karlowitz and Halasz left off. The purpose of this article is to review this work and to summarize the current understanding of nonequilibrium ionization, as applied to MHD power generation. This is an important qualification, since the literature of nonequilibrium ionization taken in the broad sense is vast indeed. It includes the literature of gas-discharge physics and low-density plasma physics. As we shall show, however, the nonequilibrium plasmas of interest in MHD power generation are much simpler than those of lower density. The essential point is that the behavior of the high-density plasma is controlled by volumetric phenomena, not by surface phenomena. This distinguishes it particularly from the glow-discharge plasmas so thoroughly studied in the past. There are a number of ways of producing a homogeneous plasma with nonequilibrium ionization. To name a few: there are the methods of photoionization, electron-beam ionization, radio-frequency excitation, radioactive decay, and, finally, nonequilibrium ionization due to (d.c.) Joule heating. It is the last of these methods which will occupy most of the present discussion. Many of the conclusions to be drawn, however, can be extended to the other methods. This follows from the fact, to be demonstrated below, that the general question of the feasibility of producing nonequilibrium ionization can be separated into two fairly distinct parts. There is first the question of the behavior of the nonequilibrium plasma, given that a certain input power is provided to maintain the electrons out of equilibrium with the gas. The question can be posed as follows: 1) What is the connection between the local electrical conductivity of the gas and the local (dissipative) pbwer density? We do not have a detailed theoretical answer to this question, which is very complex, but we do have a completely satisfactory engineering answer, which applies to most of the methods of ionization mentioned previously. The second question is as follows: 2) Can the local (dissipative) power density and gas composition be so chosen in a practical MHD generator configuration as to satisfy the conditions required to produce nonequilibrium ionization? This second question is in many ways more complex and difficult than the first, and the problem is different for each of the ionization schemes mentioned previously. We are certainly farther from having a satisfactory answer, although recent progress in the application of the scheme utilizing the natural d.c. Joule heating of the electrons is encouraging. The following discussion will be divided into two parts, The first, entitled Nonequilibrium Plasma Behavior, will deal with question 1; the second, entitled Generator Characteristics, will deal with question 2. Most of the discussion will apply to alkali-metal seeded noble gases, since these have received the greatest attention. Some work has been done on wet alkali-metal vapors, however, and these results will be mentioned briefly. A serious attempt has been made to ensure that the discussion is correct both historically and technically. The