Two scientific teams have recently engineered the H5N1 virus to make it readily transmissible between ferrets. Given that ferrets are considered the most reliable animal surrogate for human influenza infection, the newly engineered H5N1 strain is probably transmissible between humans as well. The potential consequences of an engineered human-transmissible H5N1 strain are stunning. Although seasonal flu infects as much as 20% of the world's population—more than 1 billion persons—each year, only a small fraction of those with seasonal flu dies, most often the oldest, youngest, and sickest. If the newly engineered strain were to escape the laboratory (either by design or by accident) and spread as widely as seasonal flu with anywhere near the current confirmed H5N1 human case-fatality rate, it could endanger the lives of hundreds of millions of persons. The possible benefits of this work do not justify taking such risks. As clinicians, we have a stake in this issue with our responsibilities for the diagnosis and treatment of influenza. We embrace the principle of free and open exchange of scientific information, but we also believe in the principle of “first, do no harm.” These 2 principles have come into a moment of rare conflict. It seems most reasonable and prudent to request that the involved scientific community and its institutions exercise restraint by restricting dissemination of the experimental results and discontinuing work on the engineered H5N1 strains. If a highly compelling case is made for continued work on this strain despite the risks, the work should be controlled and should merit the greatest scrutiny.