•Experiments, models and nanogranitoids are used concurrently and comparatively.•Water-fluxed melting does not produce a specific geochemical signature in melt.•Granite compositions are produced at low- to medium-P, whatever the fluid regime is.•The abundance of peraluminous granites does not imply a dry nature of orogenic crust. The diverse fluid regimes during melting of the metasedimentary crust have been often discriminated on the basis of the composition of anatectic granitoids, with granites indicating fluid-absent melting conditions and trondhjemitic compositions suggesting the addition of external water in the source region. The lack of abundant metasedimentary-derived trondhjemites in the geological record is supposed to prove the minor role of water-fluxed melting in the crustal maturation. In terms of trace elements, instead, Rb, Sr and Ba contents and their ratios have been commonly used to discriminate dehydration vs. water-fluxed melting scenarios. Here I show that reconciling results of melting experiments, thermodynamic modeling and nanogranitoid study brings out a different picture. Equilibrium thermodynamics cannot properly reproduce melt compositions of the selected benchmark experiments, with the latter having trondhjemitic compositions mainly for the metastable behavior of muscovite during laboratory runs. The formation of sufficient volumes of extractable trondhjemitic melts is related to high pressure melting conditions (≥8 kbar at 700°C and ≥11 kbar at 750°C) or K-poor bulk rock compositions, rather than to the only presence of water. At low- to medium-pressure, crustal melts are granites in composition, whatever the fluid regime is. It is inferred that the abundance of anatectic peraluminous granites (compared to metasedimentary-derived trondhjemites) does not imply a dry nature of the orogenic crust. Likewise, the use of LILE (Rb, Sr and Ba) signatures may lead to erroneous conclusions on the fluid regime of the deep continental crust.