We investigate the surface Rashba effect for a surface of reduced in-plane symmetry. Formulating a k·p perturbation theory, we show that the Rashba splitting is anisotropic, in agreement with symmetry-based considerations. We show that the anisotropic Rashba splitting is due to the admixture of bulk states of different symmetry to the surface state, and it cannot be explained within the standard theoretical picture supposing just a normal-to-surface variation of the crystal potential. Performing relativistic ab initio calculations we find a remarkably large Rashba anisotropy for an unreconstructed Au(110) surface that is in the experimentally accessible range. Metallic surfaces often exhibit Shockley-type surface states located in a relative band gap of the bulk band structure, and forming a two-dimensional electron gas. One of the most intriguing manifestation of spin-orbit coupling (SOC) at surfaces is the splitting of these surface states, known as Rashba splitting [1,2]. Such Rashba splitting was observed via photoemission by LaShell et al. [3] for the L-gap surface state at Au(111) and explained theoretically in terms of a tight-binding model [4] and ab initio electronic structure calculations [5,6], but several studies of the Rashba splitting were published in recent years on Bi(111) and Bi/Ag(111) [7][8][9], as well as on Bi x Pb 1−x /Ag(111), where atomic Bi p-orbitals lead to a more pronounced spin-orbit splitting [10][11][12][13].Describing and controlling the Rashba splitting of surface states is crucial for spintronics applications. The famous Datta-Das transistor relies on the electric tuning of the Rashba splitting [14] and the Rashba splitting is responsible for the spin Hall effect in two dimensions [15] and the anomalous Hall effect [16] as well.The simplest way to understand the origin of the Rashba effect is to take nearly free electrons, confined by a crystal potential, V (r) = V (z), and having a planewave-like wave function, ψ s,k (r) = e ikr φ (z) χ s , with χ s some spinor eigenfunctions, and k the momentum parallel to the surface. The crystal potential V (z) obviously produces an electric field, E, perpendicular to the surface, which, in the presence of spin-orbit interaction leads to the following spin-orbit term in the effective Hamiltonian,called Rashba-Hamiltonian. In Eq. (1), σ i denote the Pauli matrices andis the so-called Rashba parameter. The eigenvalue problem can then easily be solved, resulting in a splitting of the spin-degeneracy of the surface states, ε ± (k) = 2 2m * k 2 ±α R |k| , with m * the effective mass of the surface electrons [4,6]. Clearly, the above dispersion is isotropic in k-space, hence we term it as isotropic Rashba splitting. Although real systems cannot be described in terms of free electrons, and for quantitative estimates of α R the atomic structure of the potential needs be taken into account [8], the structure of the Rashba interaction, Eq. (1), is very robust for surfaces of high point-group symmetry such as C 3v or C 4v [19].The situation is, howe...