A state estimation method for 4WD vehicle was demonstrated by measuring lateral tire forces and longitudinal tire forces. Based on 3-DOF vehicle model, a state observer was realized using extended Kalman filter technique. The observer provided several advantages that the vehicle velocity and sideslip angle estimates could be robust with variance of wheel slip ratio and road friction under extreme adherence condition, such as split friction road. The paper demonstrated the appropriateness of this observer by slalom simulation test using veDYNA software. Simulation results were compared with traditional nonlinear state observer. The comparisons indicated that the longitudinal velocity and sideslip angle estimates of proposed observer both approached the true value very well. The maximum errors were no more 0.6km/h and 0.2°, comparing to 25km/h and 3° of the maximum errors of traditional nonlinear state observer. The proposed estimation method was an effective self-adaptation approach to wheel slip ratio and road friction condition. 