The development of autonomous or semi-autonomous car technology is attracting much attention in recent years. An important aspect of this research is automatic identification of road surfaces, since adjustments can be made to improve the safety of the car. This work introduces a multi-sensor road surface identification system that considers features from four different kind of sensors: microphones, accelerometers, speed signals, and handwheel signals. The system was tested on a real dataset captured using a converted passenger car driving on a closed course on three road surfaces: smooth flat asphalt, cobblestones, and stripes. Two dimensionality reduction methods were considered: feature ranking and principal component analysis. In addition, three classifiers were considered: linear discriminant analysis, support vector machines, and random forests. The best performance was obtained by random forests after feature ranking, with an accuracy of 96.52%.