In an effort to investigate the effects of relative roughness Rn and mean particle size D50 on interface shear behavior between sand and steel, a series of monotonic interface were undertaken using a modified direct shear apparatus on various types of sand (having three different mean particle sizes (samples A [D50 = 1.590 mm], B [D50 = 1.020 mm], and C [D50 = 0.638 mm]) and steel plates (having five levels of relative roughness Rn varying from 0 to 2)). Various mechanical responses from the interface shear test, such as shear stress, vertical deformation, and friction angle, were evaluated through comparisons with the trends observed in the pure soil shear test. Similar shear behavior is found among the interface shear tests having different levels of Rn. The maximum vertical deformation of interface shear test is less than that seen in the pure soil test, and the rough interface shear test also exhibits weaker shearing resistance than the pure soil shear test does. Furthermore, there exists a critical value of relative roughness Rcr whose value is between 0.5 and 1 in the present study, such that the peak shear stress or friction angle φ can no longer be readily enlarged when Rn exceeds Rcr. Moreover, the friction angles φ of samples A (D50 = 1.590 mm), B (D50 = 1.020 mm), and C (D50 = 0.638 mm) are approximately equal for any given Rn, even though the specimens having different D50 were used in the interface shear test. However, for the pure soil shear test, φ apparently increases with D50, indicating that D50 directly affects φ obtained from the pure soil test. φ obtained from the interface shear test may be not of direct interest to D50 but could depend on the coupling factor characterized by Rn.