How and when present-day massive early-type galaxies built up and what type of evolution has characterized their growth (star formation and/or merging) still remain open issues. The different competing scenarios of galaxy formation predict much different properties of early-type galaxies at z > 1. The “monolithic” collapse predicts that massive spheroids formed at high redshift (z > 2.5–3) and that their comoving density is constant at z < 2.5–3 since they evolve only in luminosity. On the contrary, in the hierarchical scenario massive spheroids are built up through subsequent mergers reaching their final masses at z < 1.5 [3,5]. As a consequence, massive systems are very rare at z > 1, their comoving density decreases from z = 0 to z ~ 1.5 and they should experience their last burst of star formation at z < 1.5, concurrent with the merging event(s) of their formation. These opposed predicted properties of early-types at z > 1 can be probed observationally once a well defined sample of massive early-types at z > 1 is available. We are constructing such a sample through a dedicated near-IR very low resolution (λ/Δλ≃50) spectroscopic survey (TNG EROs Spectroscopic Identification Survey, TESIS, [6]) of a complete sample of 30 bright (K < 18.5) Extremely Red Objects (EROs).