The origin of the energy of the stars as being due to the fusion of light nuclei was identified by Eddington in 1920. In the late 1940s, work started, aimed at creating suitable conditions in the laboratory to develop a major new energy source for the world. The two radically different approaches depend either on the inertia of a small dense hot pellet or on the use of magnetic fields to control and contain the motion of charged particles. This paper concentrates largely on the latter ‘magnetic confinement’ history leaving ‘inertial confinement’ to other contributors. Early work saw the invention of a wide variety of magnetic–field geometries for the confinement of hot ionized gases (plasmas). By a process of natural selection, this has been almost entirely reduced to so–called ‘closed toroidal systems’, such as the stellarator, tokamak and reversed–field pinch. Effort has been concentrated on the tokamak, culminating in the present position where ‘breakeven’ has been achieved in large machines. An improvement by a further factor of five to ten in the key triple–product parameter (nτT) is needed to reach the ignition point where the reaction becomes self–sustaining. New machines designed to achieve this, and in one case (International Thermonuclear Experimental Reactor) to demonstrate the technical feasibility of an ultimate power–generating system, are under consideration. A course of action in this direction is recommended.