During the last few decades interest in negative-hydrogen ion sources has been directed mainly toward synchrotron and other particle accelerator applications, with emphasis on high current densities delivered for short pulses. But within the last several years there has been an awareness in the magnetic fusion program of the future need for negative ions as a means for generating high energy neutral beams, beams with energies above a few hundred keV. Negative ions seem to be the only effective intermediary for efficiently producing such beams. Although methods for generating negative ion beams have relied upon synchrotron concepts, the requirements for fusion are very different: here one is interested in more moderate current densities, up to 100 m A cm−2, but with continuous operation. Proposed source modules would accelerate of the order of 10 A of beam current and deliver several megawatts of beam power. Both H− and D− beams are being considered for application in different reactor systems. The conceptualization of negative ion sources is now in a very volatile stage. But of the great variety of proposals that have been offered to date, three general areas appear ready for development. These are: first, the double charge exchange method for converting a positive ion beam into a negative ion beam; second, electron-volume processes wherein low energy electrons interacting with molecular species lead to negative ion products via dissociative attachment or recombination; and third, generation of negative ions in surface interactions, principally via desorption and backscattering. Both our qualitative and our quantitative understanding of these processes diminishes as one proceeds from the first through the third. The physics of these three methods is considered in detail.