Many issues arise for an Industry statistician when a potential predictive biomarker is identified for a drug under development. This talk will focus on two scenarios: 1) introducing the marker prior to start of the registrational Phase 3 trial(s), and 2) dealing with markers discovered after approval. A practical study design for internal decision making around the usefulness of the biomarker and need for refinement will be described, with oncology examples. Design and analysis considerations for supporting parallel regulatory submissions for both the NDA/BLA and the companion diagnostic kit PMA will be discussed. With respect to markers discovered after approval, challenges related to label change and modification of ongoing trials will be illustrated with examples from experiences with EGFR antagonists and K-ras.