The classic stereoscopic image quality assessment (SIQA) methods usually have the problem of low prediction accuracy,and their feature extraction process is time consum ing.To address these problems,this paper proposes a new no-reference SIQA algorithm based on synthe tic coefficients of unmatched subbands (SCUSs).First,we employ the steerable pyramid to perfo rm image decomposition over multi-scale and multi-orientation to generate multiple wave let subbands for left and right views.Then,the corresponding wavelet subbands,which are at th e same scale and the same orientation of the left and right images,are combined into SCUSs witho ut using disparity map.After that,the distribution statistics of SCUSs,the correlation features between SCUSs at the same scale and neighboring orientations,and the correlation features between SC USs at the same orientation and neighboring scales are extracted.Finally,we utilize the stati stics extracted from SCUSs to train a support vector regression model to evaluate the quality of dist orted images. Experiment results on LIVE 3D and Waterloo IVC 3D databases show that the propos ed algorithm achieves higher prediction accuracy on both symmetrically and asymmetrically dis torted images than the state-of-the-art SIQA algorithms.Meanwhile,since the SCUSs are co mbined without matching the left and right views by disparity map,the proposed algorithm can b e implemented with high efficiency.