In the study of phonation in odontocete cetaceans and particularly that of echolocation by means of sonar, a great number of conflicting hypotheses have been advanced regarding the correlation of sound production with the many anatomical features that are to be found in the upper respiratory tract. The present paper follows a new approach via finding out if, and how, the various structures had evolued from the condition in the relatively primitive River Dolphins to the more highly specialized pelagic species. An introduction to the general anatomy of the upper respiratory tract is given, relating it to work previously published on this subject. A historical review of the various theories of sound production establishes the essence of the existing controversies. One or more specimens of a representative number of species from each family have been dissected and described briefly. The species have been arranged approximately in a series of phylogenetic stages of the upper respiratory tract. It was found that there was a gradual diminution in relative size of the more superficial air chambers and an increase in the relative volume of the deeper, premaxillary diverticula. This seemed to indicate that the site of sound production is situated in the lower posterior narial tract, and that the sound is transmitted by the bones forming the rostrum. Radiographic examination of the bones of the rostrum of skulls in various stages of development was undertaken to ascertain the most favourable sound paths. It was concluded that the sonar pulses at least were produced by the larynx, which does not vary greatly in structure throughout the suborder.