This paper proposes a photo composition classification method based on fuzzy logic. For many image-based applications and processing, such as image enhancement, image segmentation, image retrial, and aesthetic evaluation, all need to meet a variety of photo compositions. The performance of the processing will be better than assumes all photos with the same photo composition for the same treatment. We develop a classification method that can be applied in some fields based on characteristics of the photo composition. The proposed Mamdani fuzzy system combines with the image features and a set of fuzzy rules. The classification result for a photo will provide the numerical information to identify the most likely photo composition. The experimental results show that the proposed method can effectively classify the photo composition as the test result by the human experience.