At what point, and under what circumstances, should trial participation be allowed to proceed without some discussion of consent? And how extensive should that discussion be? Phil Galtry and Allen Kindman of Quintiles demonstrate how this is much more than an isolated question. In the US and elsewhere, comparative effectiveness research in many fields of practice depends upon how these questions are answered. And the most extreme example of randomised investigation without informed consent occurs in trials of cardiopulmonary resuscitation for out-of-hospital cardiac arrest where, for obvious reasons, informed consent is impossible.