We discuss, in the context of the single star scenario, the nature of the progenitors of Red Supergiants (RSG), of Luminous Blue Variables (LBV) and of Wolf-Rayet (WR) stars. These three different populations correspond to evolved phases of Main-Sequence (MS) OB stars. Axial rotation and mass loss have a great influence on massive star evolution in general and more specifically on the durations of these different phases. Moderate rotation and mass loss, during the MS phase, favor the evolution towards the RSG stage. Fast rotation and strong mass loss during the MS phase, in contrast, prevent the star from becoming a RSG and allow the star to pass directly from the OB star phase into the WR phase. Mass loss during the RSG stage may make the star evolve back in the blue part of the HR diagram. We argue that such an evolution may be more common than presently accounted for in stellar models. This might be the reason for the lack of type IIP SNe with RSG progenitors having initial masses between 18 and 30 M_☉. The LBVs do appear as a possible transition phase between O and WR stars or between WNL and WNE stars. Fast rotation and/or strong mass loss during the Main-Sequence phase prevent the formation of LBV stars. The mechanisms driving the very strong ejections shown by LBV stars are still unknown. We present some arguments showing that axial rotation together with the proximity of the Eddington limit may play a role in driving the shell ejections. Rotation and mass loss favor the formation of Wolf-Rayet stars. The fact that WR stars and RSGs rarely occur in the same coeval populations indicates that the mass range of these two populations is different, WR stars originating from more massive stars than RSGs. Single star evolution models predict variations with the metallicity of the number ratios of Type Ibc to Type II supernovae, of Type Ib to Type II and of Type Ic to Type II, which are compatible with observations, provided that many stars leaving a black hole as a remnant produce an observable supernova event.