Are optically-selected QSO catalogs biased ?

A thorough study of QSO-galaxy correlations has been done on a region close to the North Galactic Pole using a complete subsample of the optically selected CFHT/MMT QSO survey and the galaxy catalog of Odewahn and Aldering (1995). Although a positive correlation between bright QSOs and galaxies is expected because of the magnification bias effect, none is detected. On the contrary, there is a significant (>99.6%) anticorrelation between z<1.6 QSOs and red galaxies on rather large angular distances. This anticorrelation is much less pronounced for high redshift z>1.6 QSOs, which seems to exclude dust as a cause of the QSO underdensity. This result suggests that the selection process employed in the CFHT/MMT QSO survey is losing up to 50% of low redshift z<1.6 QSOs in regions of high galaxy density. The incompleteness in the whole z<1.6 QSO sample may reach 10% and have important consequences in the estimation of QSO evolution and the QSO autocorrelation function.