The paper establishes tight lower bound for effective conductivity tensor K * of two-dimensional threephase conducting anisotropic composites and defines optimal microstructures. It is assumed that three materials are mixed with fixed volume fractions and that the conductivity of one of the materials is infinite. The bound expands the Hashin-Shtrikman and Translation bounds to multiphase structures, it is derived using a combination of Translation method and additional inequalities on the fields in the materials; similar technique was used by Nesi (1995) and Cherkaev (2009) for isotropic multiphase composites. This paper expands the bounds to the anisotropic composites with effective conductivity tensor K * . The lower bound of conductivity (G-closure) is a piece-wise analytic function of eigenvalues of K * , that depends only on conductivities of components and their volume fractions. Also, we find optimal microstructures that realize the bounds, developing the technique suggested earlier by and Cherkaev (2009). The optimal microstructures are laminates of some rank for all regions. The found structures match the bounds in all but one region of parameters; we discuss the reason for the gap and numerically estimate it.Keywords: multimaterial composites, optimal microstructures, bounds for effective properties, structural optimization, multiscale the periodicity cell Ω. The computed value of J allows for computation of the add outer bound of G-closure, as discussed in Sections 2.2 and 2.3. The matching microstructures (minimizing sequences) are found by a different technique that was introduces in (Albin et al, 2007a, Cherkaev, 2009 and is described in Sections 3.3 and 6; by assumption, optimal structures are laminates of some rank. The effective properties of the structures form the inner bound of G-closure. When the outer and inner bounds coincide, they are exact and the G-closure is determined. We show that our bounds are exact in all domains of parameters but one. In the last domain, we estimate the gap between the outer and inner bounds. Remark 1.1 The complementary upper bound can be established by a solution of a dual problem, in which conductivity ki are replaced by resistivity ρi = 1/ki. In the considered problem, one of the component is a superconductor (k3 = ∞) which makes the dual bound trivial -the effective resistivity can be arbitrary large, or K −1 * ≥ 0. The obtained results allows for the upper bound determination for the G-closure of materials with conductivities k1 = 0 < k2 < k3 < ∞.Bounds The problem of exact bounds has a long history. It started with the bounds by Voigt (1928) and Reuss (1929), called also Wiener bounds or the arithmetic and harmonic mean bounds. The bounds are valid for all microstructures and become in a sense exact for laminates: One of the eigenvalues of K * of a laminate is equal to the harmonic mean of the mixed materials' conductivities, and the other oneto the arithmetic mean of them. The pioneering paper by Hashin and Shtrikman (1963) found the bounds and the matchi...