In order to improve the reliability of vehicle running state estimation, a vehicle state fusion estimation method based on multi-model observer error compensation and iteration was proposed. A strong-tracking filter estimation algorithm was presented for vehicle state estimation based on three-degree-of-freedom vehicle dynamics model, meanwhile, using the coupling relationship of four wheel speed, a ridge estimation algorithm for vehicle state estimation was designed considering the influence of data disturbance and ill-conditioned matrix. To further improve the reliability of estimation system, an estimation strategy with the error compensation and iteration between the dynamic-model-based observer and the kinematic-model-based observer was developed, a fuzzy controller was designed which was used to judge the weight of strong tracking filter and ridge estimator according to the real-time pseudo measurement value of sideslip angle and longitudinal slip rate, and then estimation performance was improved by iteration and fusion of the closed-loop estimation system. The results of the simulation and road test showed that the proposed vehicle state fusion estimation method can integrate the advantages of strong tracking filter algorithm and ridge estimation algorithm, dynamically adjust the weight coefficients of strong tracking filter and ridge estimation results according to the vehicle longitudinal slip ratio and sideslip angle, guarantee the estimation accuracy and synchronously improve the adaptability of the estimation system under multiple conditions.