This paper proposes a binaural-centered mode-matching (BCMM) method that performs sound field reproduction with two reproduction points, one at each ear. A sound field reproduced by higher-order Ambisonics converges to the target field around a sweet spot whose size is inversely proportional to the frequency when the truncated order of the spherical harmonic expansion is constant. By contrast, BCMM translates the spherical harmonic coefficients on the basis of the addition theorem to realize two reproduction points at both ear positions. The BCMM method thereby avoids degradation of reproduction due to a smaller sweet spot at higher frequencies, as occurs with the conventional method, and thereby, leads to an accurate reproduction at higher frequencies. A comparison of numerical simulations of the binaural signals obtained when rendering with BCMM and conventional methods shows that, compared with the conventional method, the BCMM method improves the reproduction accuracy.