Mobile mapping systems usually include cameras designed for 360° imaging and laser scanning point cloud coloring. However, the multi-camera systems are rarely optimized for producing photogrammetric image-based point clouds in road environments. In this study, we built a mobile 5-camera system and assessed its performance in determining the 3D geometry of road surfaces. The evaluation was carried out in two parts: First, we examined how driving speed ranging from 3 to 20 km/h affect the quality of the point cloud produced by the mobile multi-camera system. We compared this data to reference measurements of road surface samples obtained using a laboratory-grade structured-light scanner. Second, we compared the point cloud produced by the mobile multicamera system to that generated by a terrestrial laser scanner from a 10-meter single lane road section. In the case of the driving speed tests, the point cloud comparisons resulted an average 3D distance from 0.09 mm to 0.31 mm, and a standard deviation from 0.29 mm to 0.50 mm. On the road section, the average 3D distance between the points clouds was 0.97 mm, with a standard deviation of 0.59 mm. These results demonstrate the capability of the mobile multi-camera system in 3D reconstruction of road surfaces and encourage further research into the feasibility of multi-camera system configurations for studying road surface quality parameters and identifying the dimensions of road damages.