This paper illustrates the appropriateness of computer graphics for the creation of realistic representations of products that are not (yet) available, for the benefit of market research applications, particularly concept tests. An experiment is conducted to study the effect of the realism of computer-created product representations on the validity of consumer evaluations. We varied the degree of realism of representations of a shaver and a picture telephone. These representations as well as the actual products are evaluated by samples of respondents. The validity of consumer evaluations increased with an increasing degree of realism, particularly for the picture telephone. That means that it may be worthwhile to invest in additional degrees of realism to get more valid consumer evaluations. Particularly, the amount of detail of the products form affected the validity of consumer evaluations. Investments in additional CAD options, such as a ray tracer or a mapper to provide a more natural impression of the products' material properties, tended to be less relevant. The improvement in validity was small in comparison with the costs. Investments in a modeller that supports an efficient and effective description of the product's form was considered to be more relevant than investments in additional CAD options, such as ray tracing and mapping.