In the last few years, many supermassive black holes have been discovered in the nuclei of nearby galaxies. Their masses are strongly correlated with masses and luminosities of the host galaxies suggesting a tight link between BH growth and galaxy evolution. In order to assess the origin of this link and the buildup of BH-galaxy relations it is mandatory to measure BH masses at all redshifts. Direct measurements based on the kinematics of gas and star are unfortunately limited to very nearby galaxies. The only possibility is therefore to use the so-called virial BH mass estimates which consist of combining continuum luminosity and broad line width measured in the spectra of Seyfert 1 galaxies and quasars. In this talk, I will briefly overview virial BH mass estimates, discuss their reliability and the available determinations of the BH galaxy relations at high redshift present in the literature. I will then concentrate on the effect of radiation pressure which has been so far overlooked. BH mass estimates neglecting the effect of radiation pressure can be severely underestimated especially in objects emitting close to the Eddington limit. I will show how, taking into account this effect, several "puzzles" like that of small BHs in Narrow Line Seyfert 1 galaxies are solved. I will then conclude by discussing new calibrated relations for high-z MBH estimates and the redshift evolution of BH-galaxy relations.