A method for estimating an image size after compression suitable for an electronic apparatus is provided. In the invention, a raw image is received, and the raw image is sampled for obtaining a sampling image. The sampling image is compressed at a plurality of testing parameters of compression respectively for obtaining a plurality of testing sizes after compression. An estimated math model is established according to the testing parameters of compression and the testing sizes after compression. A target parameter of compression corresponding to a target size after compression is calculated through the estimated math model.