It is shown that the claims that density functional theory (DFT) can handle orbitally degenerate states are ungrounded. The constraint search formulation of DFT allows one to determine a set of densities and eigenvalues for the degenerate term that, however, are neither observables, nor can they be used to solve the system of coupled equations for the nuclear motions to obtain observables, as in the wave function presentation. A striking example of the failure of the existing versions of DFT to describe degenerate states is provided by the Berry phase problem: the strong dependence of the results on the phase properties of the electronic wave function that are smeared out in the density formulation. The solution of the Jahn-Teller E-e problem illustrates these statements. For nondegenerate states with the full wave function taken in the adiabatic approximation as a product of the electronic and nuclear parts, the formulation of DFT is rigorous if and only if the dependence of the electronic wave function on nuclear coordinates is ignored. This lowers the accuracy of the results, in general, and may lead to erroneous presentation as in the case of molecular systems in strong magnetic fields. © 1997 by John Wiley & Sons, Inc.