This paper explores the hypothesis that the optimal perceptual weights for pooling in perceptual image quality assessment (IQA) algorithms should be proportional to local information content, estimated in units of bits using advanced statistical models of natural images. The authors conducted extensive studies using six publicly available subject-rated image databases and found three key findings: 1. Information content weighting consistently improves the performance of IQA algorithms. 2. Surprisingly, the widely criticized peak signal-to-noise-ratio (PSNR) can be converted into a competitive perceptual quality measure when combined with information content weighting. 3. The best overall performance is achieved by combining information content weighting with multiscale structural similarity measures (SSIM). The proposed method is based on a Gaussian scale mixture (GSM) model of natural images, which allows for the computation of local information content in bits. The information content weighting is then used to adjust the pooling stage of IQA algorithms, leading to significant performance improvements. The authors also demonstrate that the visual information fidelity (VIF) algorithm can be reinterpreted within the framework of information content weighting. Extensive tests on various image databases show that the proposed information content weighted SSIM (IW-SSIM) algorithm achieves the best overall performance, outperforming other state-of-the-art algorithms.This paper explores the hypothesis that the optimal perceptual weights for pooling in perceptual image quality assessment (IQA) algorithms should be proportional to local information content, estimated in units of bits using advanced statistical models of natural images. The authors conducted extensive studies using six publicly available subject-rated image databases and found three key findings: 1. Information content weighting consistently improves the performance of IQA algorithms. 2. Surprisingly, the widely criticized peak signal-to-noise-ratio (PSNR) can be converted into a competitive perceptual quality measure when combined with information content weighting. 3. The best overall performance is achieved by combining information content weighting with multiscale structural similarity measures (SSIM). The proposed method is based on a Gaussian scale mixture (GSM) model of natural images, which allows for the computation of local information content in bits. The information content weighting is then used to adjust the pooling stage of IQA algorithms, leading to significant performance improvements. The authors also demonstrate that the visual information fidelity (VIF) algorithm can be reinterpreted within the framework of information content weighting. Extensive tests on various image databases show that the proposed information content weighted SSIM (IW-SSIM) algorithm achieves the best overall performance, outperforming other state-of-the-art algorithms.