This paper gives the first account of the larval cranial anatomy of either of the genera of Liopelmid frogs. A single, partly grown larva of Ascaphus truei, Stejneger, has been studied in transverse sections and in two-dimensional reconstructions. Its chondrocranium, jaws, gill arches, and head muscles are described and figured. Comparisons are made throughout with similar structures of Urodeles and certain other frogs, particularly Discoglossus pictus and Kana temporaria. A summary of the characters which Ascaphus shares with the Urodeles is given on pp. 175-7 and with Discoglossus on pp. 177-8. The reader is referred to these lists as an important part of this summary. Noble (1931, &c.) considers Ascaphus (with Liopelma) to be one of the two most primitive living frogs. The findings of this paper are in full agreement with this view. Thus larval Ascaphus is shown to be a persistently primitive ‘link-animal’ whose cranial structures, in almost every case, differ from those of other frogs--often radically--and throw much light on the evolution of the modern-type frog tadpole from the unknown (larval) ancestor. Ascaphus is shown to have more characters in common with the Urodeles than any other frog larva yet described. Most of these are probably a simple retention of an ancestral Amphibian plan which led on to the frogs and Urodeles (contrast the writings of Holmgren and Säve-Söderbergh). Others seem to link these two orders even more closely together. Such are: (1) The presence of ‘urobranchial’ prongs on the basibranchial copula and the attachment to them of Subarcuales obliqui and Eecti cervicis muscles; (2) the presence of a pair of Branchio-hyoideus externus muscles and other similarities of the musculature. The relationship of Ascaphus to the Gymnophiona is far less marked. Ascaphus, however, has remained more primitive than the present-day Urodeles by retaining: (1) a? Vth gill bar, with its Subarcualis rectus and S. obliquus muscles, and (2) four pairs of S. obliqui muscles instead of two. In these points it is, in fact, the most primitive living tetrapod. Ascaphus is, however, somewhat specialized in relation to a sucker mechanism and to a peculiar method of larval progression which it employs. These have led to an exaggerated autostyly of the palatoquadrate which has developed an additional fusion to the anterior tip of the auditory capsule; to a rigid fusion of the central part of the supra-rostral system to the skull; to the general heavy build of the head cartilages and to the great size of several of the mandibular and hyoid muscles; to a general consolidation and widening of parts of the hyobranchial apparatus; to a widening of the posterior jaw cartilages and the development from them of unique posterior spurs. A pre-oral mouth cavity and a long ‘posterior narial tube’ to the inner nostril are also parts of this specialization. Among the frogs Ascaphus is shown to be most nearly related to the Discoglossidae, which appear to have been derived from an ancestor with many Ascaphus-like characters. This is in further agreement with Noble's classification and is particularly true of Discoglossus pictus. Ascaphus is unique among frogs in the posterior position of its splanchnic head structures; see the list on p. 147. A forecast is made of the probable evolution of the moderntype tadpole's jaw system from that of the unknown ancestor and this is diagrammatically summed up in Text-fig. 7, pp. 158-9. Evidence is collected to show that the ‘anterior basal process’ (= commissura quadrato-cranialis anterior) is not an ethmoidal structure by origin, as has been held up to now, and consequently Säve-Söderbergh's use of it to explain an ethmoidal structure in a Stegocephalian Amphibian is criticized. An account of the muscles has been given in summary form on pp. 126 to 146 and cannot be further condensed here. But it may be noted that Edgeworth's theories (1935) of the primitive muscular content of a single branchial segment break down when applied to Ascaphus. It is now probable that a single segment could simultaneously contain a Subarcualis rectus, a S. obliquus, and a Transversus ventralis muscle. Further, Edgeworth's term ‘Transversus ventralis II’ must be changed to ‘S. obliquus II’ in the frogs and his ‘S. rectus IV’ must probably be changed to ‘S. recti IV, III, and II’ in the Urodeles.