Early nuclear power reactor concepts were generated mainly by a few theoretical physicists during the relatively few quiet hours available in 1942–1945. Reactors were thought of in terms of structures “that a plumber could put together.” A typical illustration of early thinking is presented. In spite of these primitive beginnings, the basic technical pattern of power reactor development was understood at an early date. Estimates of world energy resources are summarized. It is pointed out that the great effort to exploit nuclear energy can be justified only if it is directed toward a full utilization of uranium and thorium. Without breeding, nuclear fuels will only supply energy for a few decades in the future energy-hungry world. Recent findings on nuclear reactor stability are discussed and the value of computing machines in theoretical studies is noted. A commentary on current concepts in reactors is presented; the flux trap reactor, boiling reactors, gas-cooling, and breeders using beryllium or U238 for possible improvement of neutron economy. Direct conversion of nuclear heat into electricity is briefly reviewed. Thermocouple batteries and thermionic converters are described and some remarks are made concerning their future development.