Abstract: There are uncertainty issues of spatial directional relations, which are generated by the positional uncertainty of deterministic geographical target and the granularity of directional system. Firstly, the rough-set representation model of spatial directional relations was proposed in cone-shaped model. Secondly, knowledge capacity measure (I(R_n)) was defined to measure the classifying ability of direction on rough-set representation of cone-shaped model. Finally, introducing I(R_n) factor to the accuracy and roughness, the accuracy and roughness based on knowledge capacity were constructed to assess quantitative uncertainty of spatial directional relations caused by the positional uncertainty of deterministic geographical target and the granularity of directional system. Experiments showed that the accuracy and roughness based on knowledge capacity are efficient as quantitative evaluation index of the uncertainty for spatial directional relations arising from the positional uncertainty of deterministic geographical target and the granularity of directional system.