Nd-Fe-B sintered magnets are used for motors of hybrid or electric vehicles due to their high energy products. Dy is added to Nd-Fe-B sintered magnets to work in a high temperature environment. Although the addition of Dy decreases the magnetization of Nd-Fe-B magnets, it increases coercivity; a decrease in the amount of Dy is strongly required. Recently, Nd-Fe-B sintered magnets with a grain size of 1 μm achieved high coercivity of ∼20 kOe without the addition of Dy or other heavy rare earth elements. In this paper, the microstructure of their magnets was observed and compared to magnets with a grain size of ∼3 μm. The coercivity of magnets consisting of larger particles was 17 kOe. Microstructures were observed by the scanning electron microscope and the shapes of grains and the distribution of the Nd-rich phase were evaluated. The observation was promoted in two directions. One direction is the plane perpendicular to the magnetically aligned direction (c plane side) and the other is the side parallel to the magnetically aligned direction (c axis side). For magnets consisting of smaller particles, the shapes of grains are isotropic for the c plane side and elongated for the c axis side. The angle of minor axis prefers to be parallel to magnetically aligned direction. The distribution of the Nd-rich phase for magnets was also evaluated for both magnets. The distribution of the Nd-rich phase at triple junctions for the magnets with smaller particles becomes homogeneous compared to that for magnets with larger particles. It is considered that Dy-free magnets with high coercivity were realized by the achievement of homogeneous distribution of Nd-rich phase besides decreasing grain size.