Abstract In this paper we discuss the contribution of different astrophysical sources to the ionization of neutral hydrogen at different redshifts. We critically revise the arguments in favour of/against a substantial contribution of active galactic nuclei (AGNs) and/or Lyman break galaxies (LBGs) to the reionization of the Universe at z > 5. We consider extrapolations of the high-z quasi-stellar objects (QSO) and LBG luminosity functions (LFs) and their redshift evolution as well as indirect constraints on the space density of lower luminosity AGNs based on the galaxy stellar mass function. Since the hypothesis of reionization due to LBGs alone requires a significant contribution of faint dwarf galaxies and a LyC photon escape fraction (f esc) of the order of ∼20 per cent, in tension with present observational constraints, we examine under which hypothesis AGNs and LBGs may provide a combined relevant contribution to reionization. We show that a relatively steep faint end of the AGN LF, consistent with present constraints, provides a relevant (although sub-dominant) contribution, thus allowing us to recover the required ionizing photon rates with f esc ∼ 5 per cent up to z ∼ 7. At higher redshifts, we test the case for a luminosity-dependent f esc scenario and we conclude that, if the observed LBGs are indeed characterized by very low f esc, values of the order of f esc ∼ 70 per cent are needed for objects below our detection threshold, for this galaxy population to provide a substantial contribution to reionization. Clearly, the study of the properties of faint sources (both AGNs and LBGs) is crucial.