The evolution of high-and low-mass X-ray binaries (HMXB and LMXB) into different types of binary radio pulsars, the ‘high-mass binary pulsars’(HMBP) and ‘low-mass binary pulsars’ (LMBP) is discussed. The HMXB evolve either into Thorne-Zytkow objects or into short-period binaries consisting of a helium star plus a neutron star (or a black hole), resembling Cygnus X-3. The latter systems evolve (with or without a second common-envelope phase) into close binary pulsars, in which the companion of the pulsar may be a massive white dwarf, a neutron star or a black hole ( some final systems may also consist of two black holes). A considerable fraction of the systems may also be disrupted in the second supernova explosion. We discuss the possible reasons why the observed numbers of double neutron stars and of systems like Cyg X-3 are several orders of magnitude lower than theoretically predicted. It is argued that the observed systems form the tip of an iceberg of much larger populations of unobserved systems, some of which may become observable in the future. As to the LMBP, we consider in some detail the origins of systems with orbital periods in the range 1–20 days. We show that to explain their existence, losses of orbital angular momentum (e.g., by magnetic braking) and in a number of cases: also of mass, have to be taken into account. The masses of the low-mass white dwarf companions in these systems can be predicted accurately. We notice a clear correlation between spin period and orbital period for these systems, as well as a clear correlation between pulsar magnetic field strength and orbital period. These relations strongly suggest that increased amounts of mass accreted by the neutron stars lead to increased decay of their magnetic fields: we suggest a simple way to understand the observed value of the ‘bottom’ field strengths of a few times 10⁸ G. Furthermore, we find that the LMBP-systems in which the pulsar has a strong magnetic field (> 10¹¹ G) have an about two orders of magnitude larger birth rate (i.e., about 4 × 10^-4 yr^-1 in the Galaxy) than the systems with millisecond pulsars (which have B < 10⁹ G). Using the observational fact that neutron stars receive a velocity kick of ∼450 km/s at birth, we find that some 90% of the potential progenitor systems of the strong-field LMBP must have been disrupted in the Supernovae in which their neutron stars were formed. Hence, the formation rate of the progenitors of the strong-field LMBP is of the same order as the galactic supernova rate (4 × 10^-3 yr^-1). This implies that a large fraction of all Supernovae take place in binaries with a close low-mass (< 2.3 M⊙) companion.