Automatic recognition of facial expressions is a vital component of natural human-machine interfaces. Facial expressions convey information about one's emotional state and helps regulate our social norms by helping detect and interpret a scene. In this paper, we propose a novel face expression recognition scheme based on Haar discrete wavelet transform and a neural network classifier. First, the sample image undergoes preprocessing where noise is removed using binary image processing techniques. Then feature vectors are extracted using DWT from corresponding pixels in the image. The extracted image pixel data are used as the input to the neural network. We demonstrate experimentally that when wavelet coefficients are fed into a back-propagation based neural network for classification, a high recognition rate can be achieved by using a very small proportion of transform coefficients. Based on our experimental results, the proposed scheme gives satisfactory results.