We present a critical assessment of the promise and pitfalls of genotype versus gene function measures in cancer epidemiology studies. While both measures have pros and cons and are complementary, in terms of potential for contributing knowledge that directly leads to prevention, we argue that attention should be given to research relating functional parameters (single protein expression, functional or phenotypic assays or patterns of gene/protein expression) to disease risk. We present the theoretical and conceptual basis of why studies focusing on polymorphisms in the low-penetrance genes may be logically less fruitful for making inroads to cancer prevention than appropriately designed studies using validated functional parameters. We then substantiate these arguments with some concrete examples based on the current literature. We also discuss the limitations of including functional parameters in epidemiological studies and technical improvements required before such studies can truly fulfil their promise. Finally, we offer some specific recommendations for future research directions in this area.