Real-time classification of road surface conditions is very important for the control of a vehicle properly within its handling limits. There have been numerous attempts to estimate the road surface conditions, but there is a plenty of room for improvement, since most of the estimation methods have some robustness issues in real-world applications. The method proposed in this paper utilizes the relationship between the road surface condition and the tire cornering stiffness. Since the tire cornering stiffness varies significantly depending on the road surface condition, it can be estimated out of the tire cornering stiffness, which is measured accurately at the early state of wheel slip. In this study, normalized tire cornering stiffnesses are estimated in real time, exploiting the fact that production vehicles are generally built to show some understeering characteristics. The proposed method compares the front and rear wheel slips simultaneously, unlike other studies that focus on individual wheel slip. In this way, estimation of unmeasurable major vehicle states such as wheel sideslip angle and absolute vehicle speed can be eliminated from the entire algorithm. The experimental results show satisfactory estimates of less than 10% error and confirm the feasibility of the proposed algorithm in production vehicles without exotic extra sensors.