The self-aeration of high velocity open-channel flows is attributed to the interaction between the turbulent flow and the liquid free surface. In addition to self-aeration, drops are ejected out of the bulk liquid. It is often proposed that the drops are separated from the bulk by transverse velocity fluctuations. Drops falling back into the liquid later have been suggested to provide a possible mechanism of air entrainment. These ideas have led to the formulation of critical conditions for the inception of self-aeration. It will be shown that the conventional concepts of drop formation are inadequate to explain the ejection of drops to the heights that were observed in experiments. Furthermore, the energy of ejected drops is not sufficiently high for air entrainment. Critical conditions of self-entrainment that were previously derived are, therefore, invalid.