A large number of targeted anticancer drugs are currently under development and most of them will have a companion diagnostic linked to their use. If a diagnostic assay is developed in conjunction with a targeted anticancer drug, such an assay will later end up determining the conditions for the use of the drug after its approval. The assay then becomes a kind of ‘gatekeeper' in relation to which patients should be treated with the drug in question. This ‘gatekeeper' role implies that companion diagnostic assays must live up to the same regulatory standards and requirements known from drug development. The assays must have proven to be analytically robust and reliable and to have demonstrated clinical utility before they are routinely used in the clinic. In the drug-diagnostic codevelopment model, several ‘traditional' study designs have been used to demonstrate clinical utility. However, if we are to benefit from the increasing knowledge provided by molecular oncology, new ways should be developed to demonstrate the clinical utility of drug-diagnostic combinations. The development of such an approach will require a rethinking at different levels and is likely to include a number of ethical, regulatory and practical challenges.