In this paper, we propose a system that will automatically rate the attractiveness of a facial image by learning the user's preferences on a set of training images. Training images are collected from a database of photographs. Each image is rated according to a user's opinion of its attractiveness. The images are preprocessed to be homogeneous in size, orientation and contrast. Processing also extracts features from the facial geometry. Classification is done using feature sets that consist of either the raw images or the extracted sets of facial geometry attributes. We compare classification performance for these two types of feature sets and find about the same performance. Classification proceeds using both parametric (Gaussian) and nonparametric (K nearest neighbors) density estimation methods.