This paper develops four neural network models to help product developers work out a combination of product form elements for best matching a given product image. By applying four most widely used rules for determining the number of hidden neurons, these four models can be used to determine the value of the product image for a given combination of product form elements. An experimental study on mobile phones is conducted to evaluate the performance of these four models. The result of this study shows that there is no best rule for building the models and the performance of these models does not differ significantly. Although the mobile phones are chosen as the object of the experimental study, the approach presented is applicable to other products where a combination of form or other design elements is to be determined for matching a desirable product image.