In this article we review the extensive, often conflicting literature on teleost gastrulation and attempt to summarize the current view and place it in a phylogenetic context. Teleosts have evolved a unique pattern of gastrulation that correlates with the presence of meroblastic cleavage. Meroblastic cleavage, on the basis of its phylogenetic distribution, appears to have evolved five times in craniates and is unique in teleosts in not being associated with an increase in egg size. Two developmental transformations, occurring at different points along the actinopterygian clade, may account for the origin of the teleost mode of gastrulation: loss of the bottle cells that appear at the beginning of gastrulation as has occurred between the Ginglymodi (Lepisosteus) and Halecostomi (Amia and Teleostei), and change in the structure of the yolk, resulting in the formation of a continuous mass from the ancestral state of a yolk segregated into platelets within individual cells as has occurred between Amia and Teleostei. These early developmental changes are explored from functionalist and structuralist perspectives for possible explanations of the evolutionary patterns seen. Functionalist explanations include the osmotic advantages offered by the telolecithal egg and the impermeable enveloping layer. Structuralist explanations encompass historical factors such as phylogenetic and developmental constraints.