On self-explaining roads drivers automatically drive according to the expected behavior and speed. In order to create this type of roads, within a given road category the layout should be homogenous, whereas a remarkable difference should exist between road categories. Using a sample of 500 persons, the paper analyses, which road categories are identified and distinguished by road users. A picture sorting task was completed to find out how road users group 45 different road scenes, and how these groups correspond to the road categories according to current standards. In addition, an analysis of a questionnaire survey was used to determine how individual road scene images are grouped on the basis of chosen speeds, and how this grouping corresponds to road categories. For processing of speed data, cluster analysis was used. The result of both surveys show that road users can clearly distinguish only 4-5 road categories. The paper compares these results with the design guidelines of some countries.