This paper presented an improved backcalculation method for rigid pavement and its regression model. Based on the modeling and a thorough analysis of the traditional AREA, the index of barcentric horizontal distance was proposed to replace the AREA index, and then the radius of relative stiffness was calculated by using each sensor deflection information. The modulus of subgrade reaction was calculated to minimize the error of filedtest deflection basin. With the above two features, the improved method can theoretically solve the problem of multiple backcalculation solutions of the traditional AREA. Taking the typical structural style of rigid pavement as a study case, the best number of sensors and the reliability of improved method were comprehensively analyzed. The numerical analysis results demonstrate that the improved method can significantly improve the reliability of backcalculation and effectively handle the measured error. The fieldtest validation results using the FWD deflection basin has also proved that compared with other methods, the improved method has a stronger ability to decrease the fitting error of measured deflection. For the improved method, five times polynomial was adopted to establish the unbiased regression model, which greatly improve the efficiency of the inversion.