Abstract: In ultrasound inverse scattering imaging,in order to improve imaging quality,the approximate total field and the unknown function should iteratively be solved. Hence,the coefficient matrices and the data of the linear equation system are both contaminated by error or noise. A truncated total least squares method is proposed,in which the error or noise in both sides of the linear equation system is considered simultaneously to solve the ultrasound inverse scattering imaging problem. Simulation results show that the truncated total least squares method can improve the goodness of fit and the quality of imaging.