The activation-synthesis hypothesis of dreaming attempted to replace the Freudian theory with a brain-based model. According to this classical model, the conscious experience of dreaming involves an interplay between excitatory and inhibitory neurons in the pons. Recent research has cast doubt on this classical hypothesis by demonstrating that pontine lesions do not lead to a cessation of dreaming, implying that the pons may be unnecessary for dreaming. However, scrupulous human lesion studies are required in order to clarify the role of the pons in dreaming. Three case examples with lesions to various brainstem regions are reported in this paper to illustrate the paradox and difficulties of carrying out pontine-lesion studies.