The concept of using a predictive or selective diagnostic assay in relation to drug development goes back to the 1970s when the selective estrogen receptor modulator, tamoxifen (AstraZeneca) was developed for metastatic breast cancer. Clinical data showed that the estrogen receptor status correlated well with the clinical outcome when the patients were treated with tamoxifen. Although this datum was published more than 35 years ago, the adaption of the drug–diagnostic codevelopment model has been relatively slow, and it is only within the last decade that it has gained widespread acceptance. In this model, the drug and the diagnostic are interdependent, and if the development project proves successful, the companion diagnostic assay (CoDx) will end up determining the conditions for the use of the drug. This gatekeeper role obviously requires that the CoDx assays adhere to the same strict rules and regulations that are known from the development of drugs. For any CoDx assay, it must be documented that it is robust and reliable and that it possesses clinical utility. The article focus on some of the most important aspects of the CoDx development process with emphasis on the clinical validation and clinical utility but also other critical issues such as the biomarker selection process, determination of the cut-off value, and the analytical validation are discussed.