There is a general agreement that the M dwarfs constitute the most numerous objects in our Galaxy. Only at the faint end (Mv >= 12m) of the main sequence the debate is still open how the stellar luminosity function and mass function continue. Is there a local mass of the disk population or not. The DENIS Survey (Deep Near-IR Southern Sky Survey), a project in which some of us are directly participating, should discover a lot of new faint nearby dwarfs and contribute to a significant improvement of the stellar luminosity function in the solar neighborhood towards fainter magnitudes. However, the present estimates of the effective temperatures for individual objects with Teff < 4000K are quite poor (see Monet et al. 1992,AJ 103,638), and a large fraction of the flux is emitted in the infra-red. It would be highly desirable to receive low-resolution spectra for carefully selected stars, well-distributed along the main sequence, beyond the DENIS reddest band. As the error in absolute magnitude depends essentially on the error of the parallax determination, we selected stars contained in the Hipparcos Input Catalogue, for which very soon accurate parallaxes become available. So we will have the full advantage of combining ISO, Hipparcos and DENIS data for the proposed objects. The objects were selected from the Catalogue of Nearby Stars III (Gliese and Jahreiss, 1991, preliminary version), according to the following criteria: (i) single stars, or wide-pair with well-isolated components; (ii) not variable whenever possible; (iii) away from the galactic plane; (iv) no disturbing neighbours or future neighbours (our stars are all high-proper motion stars, check in the HST Guide-Star Catalog); (v) declination south of +2 degrees, in order to be observed with DENIS, and used as standards; (vi) in the Hipparcos Input Catalogue, except one too faint for it, but with an excellent ground-based parallax.