This paper describes a method to relate form impressions to design parameters; a mapping of image-space to design-space via a neural network solution. A survey designed by the Taguchi method was used to collect data on how altering various design parameters of an automobile affects its image. Principal Component Analysis was performed to extract correlated factors from this data and each sample included in the survey was described in this new space by calculating its factor score. A neural network was constructed and trained with these samples, which allowed a CAD-system, based on this neural network, to automatically generate 3D-models corresponding to any form impression presented to it These results show that it is possible to give a design system a 'sense' of shapes, previously restricted only to the designer.