We investigate why no hydrogen-disordered form of ice II has been found in nature despite the fact that most of hydrogen-ordered ices have hydrogen-disordered counterparts. The thermodynamic stability of a set of hydrogen-ordered ice II variants relative to ice II is evaluated theoretically. It is found that ice II is more stable than the disordered variants so generated as to satisfy the simple ice rule due to the lower zero-point energy as well as the pair interaction energy. The residual entropy of the disordered ice II phase gradually compensates the unfavorable free energy with increasing temperature. The crossover, however, occurs at a high temperature well above the melting point of ice III. Consequently, the hydrogen-disordered phase does not exist in nature. The thermodynamic stability of partially hydrogen-disordered ices is also scrutinized by examining the free-energy components of several variants obtained by systematic inversion of OH directions in ice II. The potential energy of one variant is lower than that of the ice II structure, but its Gibbs free energy is slightly higher than that of ice II due to the zero-point energy. The slight difference in the thermodynamic stability leaves the possibility of the partial hydrogen-disorder in real ice II.