When deposited on a hot solid, volatile liquids can levitate on a cushion of their own vapour, without liquid-solid contact, a phenomenon called the Leidenfrost effect. The presence of a vapour layer, that separates the liquid from the solid, has two crucial consequences for applications: (i) a dramatic reduction of heat transfer, and (ii) a reduced friction, making Leidenfrost drops extremely mobile. In this seminar, I will tackle a deceivingly simple question: what is the solid temperature necessary to observe levitation, the so-called Leidenfrost temperature. First, I will propose to understand the onset of the Leidenfrost effect as emerging from the ability of the vapour layer to levitate the liquid, through the example of drop impact on superheated substrates. I will then show the limitations of this approach, and propose a different definition of the Leidenfrost temperature as the critical point of a second order phase transition, by making an analogy with non-equilibrium systems exhibiting a directed percolation phase transition. In this framework, the Leidenfrost temperature does not stem from the hydrodynamic ability of the vapour layer to support the liquid, but it expresses the resilience of Leidenfrost drops to contact. Finally, I will discuss the implications of this new definition of the Leidenfrost temperature on heat transfer in drop impact situations.