The charge-exchange spin-dipole (SD) excitations for both (p, n) and (n, p) channels in 16 O are investigated in the fully self-consistent random phase approximation based on the covariant density functional theory. The fine structure of SD excitations in the most up-to-date 16 O( p, n) 16 F experiment is excellently reproduced without any readjustment in the functional. The SD excitations are characterized by the delicate balance between the σ-and ω-meson fields via the exchange terms. The fine structure of SD excitations for the 16 O(n, p) 16 N channel is predicted for future experiments. PACS numbers: 24.30.Cz, 21.60.Jz, 24.10.Jv, 25.40.Kv The nuclear charge-exchange excitations [1] correspond to the transitions from the ground state of the nucleus (N, Z) to the final states in the neighboring nuclei (N ∓ 1, Z ± 1) in the isospin lowering T − and raising T + channels, respectively. These excitations can take place spontaneously such as the well-known β decays or be induced by external fields such as the charge-exchange (p, n) or (n, p) reactions. They are categorized according to the orbital angular momentum transfer as allowed transitions with L = 0 and first-and second-forbidden transitions with L = 1 and L = 2, etc. Meanwhile, they are also classified by the spin's degree of freedom as the non-spin-flip modes with S = 0 and the spin-flip modes with S = 1.Among all the nuclear charge-exchange excitation modes, the spin-dipole (SD) excitations with S = 1 and L = 1 have attracted more and more attentions due to their connection with the neutron-skin thickness [2], the cross sections of neutrino-nucleus scattering [3,4], the double beta decay rates [5], and so on. Different from the famous Gamow-Teller (GT) excitations having a single spin-parity J π = 1 + component, the SD excitations are composed of three collective components with spin-parity J π = 0 − , 1 − , and 2 − . It is relatively straight forward to distinguish the orbital angular momentum transfer L by the angular distributions of double differential cross sections, but it is not trivial to resolve different J π components in SD excitations [1]. However, the resolution of these three J π components is crucial to understand the multipole-dependent effects on the neutrino-nucleus scattering [6] and neutrinoless double beta (0νββ) decays [7,8], and the strengths of nucleon-nucleon effective tensor interactions [9, 10] for understanding the evolution of the single-particle energies in exotic nuclei [11,12]. In particular, the J π = 0 − states can also serve as doorways for parity mixing in compound nuclear states [13]. Therefore, the investigation of the fine structure for SD excitations including all J π components has become one of the central issues for both experimental and theoretical nuclear physics, particle physics, and astrophysics.Charge-exchange excitations in 16 O are of particular interest in both nuclear physics and astrophysics. For instance, 16 O is the key nucleus in the waterČherenkov detector for (anti-)neutrinos providing evident s...