The problem is that we also don't know how many people might be killed by an unsafe drug. It's impossible to know in advance what could fail with a drug - hence

The problem is that we also don't know how many people might be killed by an unsafe drug. It's impossible to know in advance what could fail with a drug - hence the testing. We have no concept of the upper bound of possible deaths. If a major drug is released without enough drug trials and widely prescribed we could lose thousands. See Voixx for an example. Almost 40,000 people died [0]. This was with a comprehensive drug trial under the FDA process. We would have many more incidents like this if not for drug trials required by regulatory agencies. On the other hand national statistics on deaths caused by various diseases have been compiled for decades. We can know with certainty the upper bound of people killed per year (which in your example isn't 1000 but some fraction of that - no drug is 100% effective). Also, many drugs don't prevent deaths but merely comfort the patient or remove some symptoms. Many drugs have alternates which could be taken. Many drugs have an upper bound of deaths prevented of 0. Why should we risk any lives for a drug which cannot save any additional lives? In general humans prefer taking a known risk with known odds rather than an unknown risk with unknown odds. If an informed consumer wants to make a different choice than the FDA, nobody is stopping them from traveling to a country where the drug is already available. [0] http://www.npr.org/templates/story/story.php?storyId=5470430 http://www.npr.org/templates/story/story.php?storyId=5470430