The diffraction-limited observations of the Hinode Spectro-Polarimeter have open exciting possibilities for the analysis of the weak magnetic signals of the solar internetwork (IN). Hinode/SP has demonstrated that the IN is permeated by highly inclined fields with weak flux densities. These observations are important to solve the discrepancy between the IN field strength distributions obtained using visible and infrared ground-based measurements. Recently, it has been argued that the information contained in the Fe I 630 nm spectral region is not sufficient to provide reliable field strength values. In this contribution, radiative magnetoconvection simulations are used to generate synthetic Hinode/SP observations, in an attempt to examine the reliability of Milne-Eddington inversions applied to very high spatial resolution data. The results show that ME inversions deliver reasonably good magnetic field strengths and inclinations from Hinode/SP measurements, provided the effects caused by telescope diffraction are corrected. These effects can be accurately modeled using a local stray-light contamination. If one does not account for the reduction in polarization signals caused by diffraction, the inversion leads to field strengths that are too weak. Finally, the implications that telescope diffraction has on the average magnetic filling factors and flux densities retrieved from Hinode/SP data are discussed.