The work aims to study the no-reference quality assessment method for multiply distorted images, in order to solve the problem that the images are simultaneously distorted by multiple types of distortions in practice. First of all, the discriminative model of no-reference multiply distorted images was established by spatial and spectral entropies and singular values. Then, according to the different multiple types of distortions, the no-reference quality assessment (NR-IQA) model for multiply distorted images was established by respectively extracting such three types of different image information features as multi-dimensional spatial statistical feature, singular value variation and spatial and spectral entropies. Moreover, the scores were obtained by selecting the optimal NR-IQA model. The discrimination rate of such method for the multiple distortion types could reach up to 100%. For the multiply distorted images, involving fuzzy noise and fuzzy compression, the spearman rank-order correlation coefficient (SROCC) on the LIVEMD database were 0.9874 and 0.9916, respectively, which provided good subjective consistency. The experiment results show that, such no-reference quality assessment method for multiply distorted images is in good subjective consistency with the ocular perception.