I present progenitor luminosities (L) for a sample of 112 Type II supernovae (SNe II), computed directly from progenitor photometry and the bolometric correction technique, or indirectly from empirical correlations between progenitor luminosity and [O i] λλ6300, 6364 line luminosity at 350 d since explosion, 56 Ni mass, or absolute V -band magnitude at 50 d since explosion. To calibrate these correlations, I use twelve SNe II with progenitor luminosities measured from progenitor photometry. I find that the correlations mentioned above are strong and statistically significant, and allow to estimate progenitor luminosities to a precision between 20 and 24 per cent. I correct the SN sample for selection bias and define a subsample of 112 SNe II with progenitor luminosities between log(L/L ⊙ ) = 4.6 dex, corresponding to the completeness limit of the corrected sample, and the maximum observed progenitor luminosity of log(L/L ⊙ ) = 5.091 dex. The luminosity distribution for this subsample is statistically consistent with those for red supergiants (RSGs) in LMC, SMC, M31, and M33 with 4.6 ≤ log(L/L ⊙ ) ≤ 5.091. This supports that SN II progenitors correspond to RSGs. The conspicuous absence of SN II progenitors with log(L/L ⊙ ) > 5.1 dex with respect to what is observed in RSG luminosity distributions, known as the RSG problem, is significant at a 5.2 ± 0.5 σ level.