The authors describe a new training methodology for generative adversarial networks. The key idea is to grow both the generator and discriminator progressively: starting from a low resolution, add new layers that model increasingly fine details as training progresses. This both speeds the training up and greatly stabilizes it, allowing to produce images of previously unprecedented quality, e.g., CelebA images at 1024^2. The authors also propose a simple way to increase the variation in generated images, and achieve a record inception score of 8.80 in unsupervised CIFAR10. Additionally, the paper describes several implementation details that are important for discouraging unhealthy competition between the generator and discriminator. Finally, the authors suggest a new metric for evaluating GAN results, both in terms of image quality and variation. As an additional contribution, the paper presents a higher-quality version of the CelebA dataset.