The idea of using atomic power as the basis for a rocket engine predates the First World War, but at that time neither liquid rocket engines nor atomic power were realities, and these concepts were essentially speculations about speculations. However, by the end of World War II, both large liquid-fuel rocket engines and atomic reactors were working technology. In 1946, Douglas Aircraft performed the first formal study of the use of nuclear power for rocket propulsion, followed by a study performed in 1947 by Dr. Hsue-Shen Tsien, then at MIT. With enthusiasm for space exploration and nuclear power rising in the 1950s and into the 1960s, interest in using "nuclear thermal rockets (NTR)" reached a high pitch.

An NTR is very simple in concept. A light propellant, usually hydrogen, is pumped through a reactor core, which heats it to high temperatures, on the order of 2,500 degrees Celsius, and spews it out as a high-velocity exhaust. An NTR provides at least twice the specific impulse of LOX-RP propulsion. However, the improved mass ratio is partially cancelled by the mass of the reactor, and particularly the mass of the reactor shielding.

For this reason, many NTR spacecraft design concepts featured a crew compartment and propulsion system separated by a long truss to allow reduction in shielding mass, while still providing protection for the crew and spacecraft systems. A well-known design along such lines was the nuclear-powered "Discovery" Jupiter spacecraft used in Stanley Kubrick's meticulously detailed 1968 movie 2001: A SPACE ODYSSEY. Incidentally, the actual thrust generation system for the Discovery spacecraft was not defined, and the details more suggest a nuclear electric ion propulsion scheme, discussed in the next chapter, than an NTR.

* In reality, 2001 came and nothing of the sort ever happened. The US Los Alamos National Laboratory, run by the US Atomic Energy Commission (AEC, now the Department of Energy / DOE) performed basic studies on NTR under the ROVER program, beginning in 1955, with NASA also beginning studies of its own in the late 1950s. The first ground test of an NTR, designated "Kiwi-A", was conducted in 1959 by Los Alamos. It was strictly a proof-of concept demonstrator, not intended for flight, which is why it was named after the New Zealand kiwi, a flightless bird.

The Kiwi-A run was followed by "Kiwi-A Prime" and "Kiwi-A3" tests, featuring minor improvements, and then the "Kiwi-B" engine, which featured a developed fuel system. The last Kiwi tests were in 1964. They were followed by the "Phoebus" series of tests, which featured much more powerful engines. The first "Phoebus-1A" test was in June 1965, with "Phoebus-2" following in February 1967, and "Phoebus-2A" completing the tests in June 1968. Phoebus-2A produced an astounding 4 gigawatts of power. A scaled-down successor to Kiwi, the "Peewee", was also tested several times.

* Although these engines were test systems, not intended for flight, in the meantime NASA formed a collaboration with the AEC to build a flight-worthy engine under the "Nuclear Engine for Rocket Vehicle Application (NERVA)" program. In May 1961, US President John F. Kennedy had called for high-priority development of NTR engines as an item in the famous Moon-landing speech, accelerating work on the project.

Within a few years, the NERVA program was focusing on an NTR that could be launched as an upper stage on the big Saturn V booster, then in development. The initial test run of the engine was in September 1964. More than 20 prototype NERVA engines were built and successfully ground-tested, with power levels between 40 and 4,000 megawatts, thrust levels up to 981 kN (100,000 kgp / 220,000 lbf), and the desired high exhaust velocities. Although the engineering problems were substantial, a flight test model of a NERVA engine was under development when the program was cancelled in 1972, after the Nixon Administration cut NASA's plans for manned Mars exploration.

Of course, whatever the US did the USSR usually did. In the early 1960s the Soviet Union built a test facility named "Baikal-1" in Kazakhstan to develop an NTR, leading to initial tests of a prototype named "IRGIT" in 1978 and some further development work during the 1980s.