A specifically designed force-feedback device accurately simulated textures consisting of lateral forces opposing motion, simulating friction. The textures were either periodic trapezoidal forces, or sinusoidal forces spaced at various intervals from 1.5 mm to 8.5 mm. In each of two experiments, 10 subjects interacted with the virtual surfaces using the index finger placed on a mobile plate that produced the forces. The subjects selected their own speed and contact force for exploring the test surface. The apparatus returned force fields as a function of both the finger position and the force normal to the skin allowing full control over the tangential interaction force. In Experiment #1, subjects used an integer, numerical scale of their own choosing to rate the roughness of eight identical, varyingly spaced force ramps superimposed on a background resistance. The results indicated that subjective roughness was significantly, but negatively, correlated (mean r = -0.84) with the spatial period of the resistances for all subjects. In a second experiment, subjects evaluated the roughness of 80 different sinusoidal modulated force fields, which included 4 levels of resistance amplitude, 4 levels of baseline friction, and 5 spatial periods. Multiple regression was used to determine the relationship between friction, tangential force amplitude, and spatial period to roughness. Together, friction and tangential force amplitude produced a combined correlation of 0.70 with subjective roughness. The addition of spatial period only increased the multiple regression correlation to 0.71. The correlation between roughness estimates and the rate of change in tangential force was 0.72 in Experiment #1 and 0.57 in Experiment #2. The results suggest that the sensation of roughness is strongly influenced by friction and tangential force amplitude, whereas the spatial period of simulated texture alone makes a negligible contribution to the sensation of roughness.