Most recent advances in cancer treatments are based on molecularly targeted drugs which benefit only those whose tumors are de-regulated by somatic alteration of a target of the drug. Many of these successes used enrichment designs for clinical development rather than conventional broad eligibility clinical trials. Even in such relatively straightforward cases, however, there are challenges in co-development of a new drug with a companion diagnostic. One needs an analytically validated assay, one must decide how to measure de-regulation of the target, one needs a threshold for assay positivity, one must design phase II trials with more complex objectives that previously and consider the seamless phase II/III adaptive enrichment approach, one must judge whether an enrichment design is appropriate for the drug, and one must deal with statisticians, regulators and sponsors who are uncomfortable with new paradigms. I will discuss these issues.