Pineal evolution is envisaged as a gradual transformation of pinealocytes (a gradual regression of pinealocyte sensory capacity within a particular cell line), the so-called sensory cell line of the pineal organ. In most non-mammals the pineal organ is a directly photosensory organ, while the pineal organ of mammals (epiphysis cerebri) is a non-sensory neuroendocrine organ under photoperiod control. The phylogenetic transformation of the pineal organ is reflected in the morphology and physiology of the main parenchymal cell type, the pinealocyte. In anamniotes, pinealocytes with retinal cone photoreceptor-like characteristics predominate, whereas in sauropsids so-called rudimentary photoreceptors predominate. These have well-developed secretory characteristics, and have been interpreted as intermediaries between the anamniote pineal photoreceptors and the mammalian non-sensory pinealocytes. We have re-examined the original studies on which the gradual transformation hypothesis of pineal evolution is based, and found that the evidence for this model of pineal evolution is ambiguous. In the light of recent advances in the understanding of neural development mechanisms, we propose a new hypothesis of pineal evolution, in which the old notion 'gradual regression within the sensory cell line' should be replaced with 'changes in fate restriction within the neural lineage of the pineal field'.