Authorities have to be on the alert regarding potential causes of human rights violation. Human rights instruments require from them to take action in order to secure the rights of those within their jurisdictions. The extent of these positive obligations is not unlimited and depends on many factors, among which a risk assessment can play a significant role: was there a (real and immediate) risk of violation and has it been taken into consideration seriously enough by the authorities? In the field of the right to life, on which the paper will focus, the obligation to prevent risks (of crimes, of accidents, of damages resulting from natural disasters, etc.) is a factor of security, but it can also restraint the capacity of action of democratic authorities. I will examine how judges (especially the ECtHR) proceed when they have to evaluate risks. Do they apply methods inspired from the risk management literature or do they reason in a rather empirical or even instinctive manner?