An intensified focus on drug safety often leads to demands for more data collection prior to drug approval. Other approaches can be used, such as enhanced post-marketing surveillance, but the optimal approach must be chosen for each as as it arises. Many drug benefits and adverse effects are unlikely to become apparent before wide distribution among diverse patients. The best balance of pre- and post-approval data collection may vary by drug. The consequences of alternative strategies are complex and not always immediately apparent, so formal modeling offers the best approach to determine which strategy is optimal in each case.