QSOs are intrinsically luminous and therefore can be seen rather easily at large distances; but they are rare, and finding them requires surveys over large areas. As a consequence, at present, the number density of QSOs at high redshift is not well known. Recently, the Sloan Digital Sky Survey (SDSS) has produced a breakthrough, discovering QSOs up to z=6.43 [8] and building a sample of six QSOs with z>5.7. The SDSS, however, has provided information only about very luminous QSOs (M$_{1450} \lesssim -26.5$), leaving unconstrained the faint end of the high-z QSO Luminosity Function (LF), which is particularly important to understand the interplay between the formation of galaxies and super-massive black holes (SMBH) and to measure the QSO contribution to the UV ionizing background [16]. New deep multi-wavelength surveys like the Great Observatories Origins Deep Survey (GOODS), described by M.Dickinson, M.Giavalisco and H.Ferguson in this conference, provide significant constraints on the space density of less luminous QSOs at high redshift. Here we present a search for high-z QSOs, identified in the two GOODS fields on the basis of deep imaging in the optical (with HST) and X-ray (with Chandra), and discuss the allowed space density of QSOs in the early universe, updating the results presented in [6].