To place core-collapse supernovae (SNe) in context with the evolution of massive stars, it is necessary to determine their stellar origins. I describe the direct identification of SN progenitors in existing pre-explosion images, particularly those obtained through serendipitous imaging of nearby galaxies by the Hubble Space Telescope I comment on specific cases representing the various core-collapse SN types. Establishing the astrometric coincidence of a SN with its putative progenitor is relatively straightforward. One merely needs a comparably high-resolution image of the SN itself and its stellar environment to perform this matching. The interpretation of these results, though, is far more complicated and fraught with larger uncertainties, including assumptions of the distance to and the extinction of the SN, as well as the metallicity of the SN environment. Furthermore, existing theoretical stellar evolutionary tracks exhibit significant variations one from the next. Nonetheless, it appears fairly certain that Type II-P (plateau) SNe arise from massive stars in the red supergiant phase. Many of the known cases are associated with subluminous Type II-P events. The progenitors of Type II-L (linear) SNe are less established. Among the stripped-envelope SNe, there are now a number of examples of cool, but not red, supergiants (presumably in binaries) as Type IIb progenitors. We appear now finally to have an identified progenitor of a Type Ib SN, but no known example yet for a Type Ic. The connection has been made between some Type IIn SNe and progenitor stars in a luminous blue variable phase, but that link is still thin, based on direct identifications. Finally, I also describe the need to revisit the SN site, long after the SN has faded, to confirm the progenitor identification through the star's disappearance and potentially to detect a putative binary companion that may have survived the explosion.This article is part of the themed issue 'Bridging the gap: from massive stars to supernovae'.