This is a critical review of theories of known discrete X-ray sources. The Crab is omitted, having been dealt with in Woltjer's review. Two of the identified sources, Sco X-1 and Cyg X-2, seem to be of the same sort. A binary or gas-stream model like that of Prendergast and Burbidge, with dimension R ∼ 109 cm and density n ∼ 1016 cm−3, appears reconcilable with the observed features of these sources, though much detailed work remains to be done. Neither object is yet known to be binary. Theoretical work becomes more difficult if, as appears to be the case at least for Sco X-1, the objects are optically thick due to electron scattering; this may affect the optical and X-ray spectra.The recent searches for iron lines in the X-ray spectrum of Sco X-1 are reviewed briefly. The calculations and the energy resolution are not yet good enough to make this a dependable test of models.Several possibilities are offered for explaining the excess radio flux from Sco X-1.Other theories of Sco X-1-type sources are discussed briefly. The theory of Manley and Olbert seems a little superfluous when the gas-stream theory is still in a strong position.There are serious discrepancies between X-ray and optical estimates of the distance to Sco X-1. 21-cm measurements must also be considered. The situation is reviewed, and ways out of the difficulty are discussed.Cen X-2 seems to be like Sco X-1, but several other unidentified sources have hard spectra like the Crab. It is tempting to speculate that most of the galactic sources are supernova remnants.The extended γ-ray source in the galactic plane may be the extrapolated unresolved sum of galactic X-ray sources, as suggested by Ogelman. There are several other possibilities.M87 is the only established extragalactic source. Radio, optical and X-ray observations are summarized and graphed. A power-law extrapolation to the X-ray band is far from mandatory; nevertheless the optical flux from the jet is known to be synchrotron radiation. The time-scale difficulties in the jet are described, and several theories of the survival of the optical electrons are reviewed.Processes for producing X-rays other than thermal bremsstrahlung and synchrotron radiation are listed. These other processes are characterized by low efficiency, and are likely to be unimportant in discrete sources, though several have attracted attention with reference to the diffuse background.