The thermodynamic stability of dichromium carbonyls is investigated with density functional theory (DFT). The results demonstrate why [(mu -H)Cr-2(CO)(10)](-) has been observed while the Cr-2(CO)(11) and (mu -H)(2)Cr-2(CO)(9) structures remain unknown. The related structure [(mu -H)(2)Cr-2(CO)(8)](2-) is predicted to be stable with respect to its fragments and isolable. Homoleptic chromium carbonyl structures of the formula Cr-2(CO)(11) appear to be thermodynamically unstable with respect to dissociation to the fragments Cr(CO)(6) and Cr(CO)(5) and only slightly metastable with respect to the transition state leading to these dissociated fragments. The potential energy surface in the region adjacent to these minima appears to be very flat. In contrast, both the BP86 and B3LYP functionals predict the known [(mu -H)Cr-2(CO)(10)](-) to have significant stability with respect to the fragments Cr(CO)(5) + [Cr(CO)(5)H](-). For the B3LYP functional, the dissociation energy is 41 kcal/mol, while for BP86 it is 43 kcal/mol. A notable structural difference for [(mu -H)Cr-2(CO)(10)](-) between the two theoretical methods is that the BP86 functional predicts the Cr-H-Cr angle to be 147 degrees while the B3LYP functional predicts a linear geometry (180 degrees). Experimental structures of [(mu -H)Cr-2(CO)(10)](-) determined by neutron diffraction and by X-ray crystallography display a remarkably similar ambiguity in the Cr-H-Cr angle. Certain other differences between the B3LYP and BP86 functionals are observed in the predicted geometries, numbers of imaginary vibrational frequencies, and particular energy differences. Several subtle comparisons suggest that the BP86 method is preferable to B3LYP for this particular class of compounds.