A type of nuclear rocket propulsion proposed by NASA's Jet Propulsion Laboratory and explained on the (now non-existent) Advanced Propulsion Concepts web site. The following is a summary ,in my own words, of that article. I cannot properly cite the source because the source no longer exists.
Current nuclear rocket concepts are of the solid-core variety, which means they use a standard nuclear fission reactor to heat a working fluid. Because of material constraints, the temperature of the working fluid can't be much more than 3000 K, even though the reactor has a much higher potential. Since specific impulse is directly proportional to temperature, the specific impulse is relatively low as a result; 900 seconds seems to be the maximum.
The idea of nuclear fission fragment propulsion is revolutionary; if one were to dispense with the idea of working fluids and use the nuclear fuel itself as the exhaust, much higher specific impulses could be achieved, maybe even as high as 1 million seconds. The reason for the amazingly high specific impulse is the extremely high velocity that the nuclear fuel fragments could achieve extremely high velocities, and specific impulse is directly proportional to exhaust velocity.
No prototypes have been built for such an engine, as it is only an idea, but if it were to be built, the results would be amazing. Around 5 million times less fuel would need to be consumed for a certain change in velocity versus the standard chemical rockets in use today.