The CH··O contacts in the 3,4-bis(dimethylamino)-3-cyclobutene-1,2-dione (DMACB) crystal have been characterized through a topological analysis of its experimental and theoretical densities, derived from a multipole refinement of X-ray diffraction data and from periodic Hartree−Fock calculations, respectively. The existence or the lack of an H··O bond critical pointthat is a point through the two nuclei where the gradient of the electron density vanishesallows us to distinguish between bonded and nonbonded CH··O contacts, regardless of the value of their H··O separation. The 23 unique bonded contacts in DMACB are characterized by a large and nearly constant (∼140°) C−H−O angle, denoting the importance of the electrostatic energy contribution to such interactions. Instead, the nonbonded ones (four unique for H··O separations up to 3.0 Å) are more bent and may even be folded down to 90°, since their dominant van der Waals contribution to the interaction energy is independent of the C−H−O angle. The CH··O angular distribution observed for H··O separations greater than 2.7 Å is only apparently isotropic, since such isotropy clearly disappears when the bonded and nonbonded contacts are identified and their angular distributions separately analyzed. The Koch and Popelier criteria (J. Phys. Chem. 1995, 99, 9747) to establish H-bonds are, for the first time, applied in their entirety to a large set of CH··O contacts in a crystalline phase. The criteria are always satisfied by all of the bonded intermolecular CH··O contacts, with a single exception concerning one long bond and one of the six criteria only. The expressions proposed by Espinosa et al. (Chem. Phys. Lett. 1998, 285, 170), relating the potential energy densities at the critical point to the H-bond strengths, fail when applied to the weak CH··O interactions present in the DMACB crystal. The reasons for such a failure are outlined and new relationships are proposed. The importance of the promolecular charge distributions in defining topological properties of interest to the CH··O bonds is investigated. The criticism raised by Spackman (Chem. Phys. Lett. 1999, 301, 425) as to the lack of additional information provided by the experimental results to the description of such weak interactions is discussed. It is shown that the promolecular model yields significantly different electron density values at the critical point and in some instances even different topologies, compared to the corresponding multipole or theoretical densities. On the other hand, when the electron density topologies are the same, the values obtained from either electron density for the potential or kinetic energy density at the critical point, are very much alike.