The mechanical properties of sandy road are quite different from those of hard surface road. For vehicle control systems such as EMS (engine management system), TCU (transmission control unit) and ABS (antilock brake system), the strategies and parameters set for solid surface road are not optimal for driving on sandy road. It is an effective way to improve the mobility of all-terrain vehicles by identifying sandy road online and shifting the control strategies and parameters of control systems to sandy sets. In this paper, a sandy road identification algorithm for SUVs is proposed. Firstly, the vehicle signals, such as engine torque and speed, gear position, wheel and vehicle speed, are acquired from EMS, TCU and ESP (electronic stability program) through CAN (controller area network) bus respectively. Based on the information and longitudinal force equilibrium equation, the travelling resistance of vehicle is estimated. The hydraulic torque converter is divided into several parts to calculate the acceleration resistance instead of using the rotational inertia coefficient. Then, the sandy road identification algorithm is proposed mainly based on the travelling resistance. Finally, real vehicle tests are carried out on different road conditions. After cone index penetrometer and soil hygrometer are used to measure the sandy test fields, performances of the travelling resistance estimation method and sandy road identification algorithm are validated. The results show that the identification algorithm designed in the paper can identify the sandy terrain effectively.