Amphibian material collected by Mr F. R. Parrington from an Upper Permian locality in the Ruhuhu Valley of Tanganyika has been prepared with the aid of acid. An account of the technique adopted is given. The remains consist of three imperfect skulls, some skull fragments and about 300 post-cranial bones, most of which were contained in blocks of marly limestone. The material is assigned to a new genus and species for which the name Peltobatrachus pustulatus is proposed and a diagnosis given. The cranial material is described and a composite restoration is given. The pattern of dermal roofing bones is of the `temnospondyl' type, but the surface is ornamented with pustules instead of the normal labyrinthodont ornament. The whole skull, including supraoccipital, basioccipital and opisthotic, is heavily ossified. Parts of a jaw ramus, including a well-developed retroarticular process, are described. The dorsal vertebrae are shown to consist of single, elongate notochordal centra and free `intervertebral' neural arches. It is concluded that the dorsal vertebrae are not stereospondylous. Caudal vertebrae consisting of neural arches, centra and separate small intercentra with haemal arches are described and a reconstruction is made of the caudal region. The dorsal ribs are adapted to carry the heavy dermal armour: a complete sacral rib and forked caudal ribs are also described. Imperfect remains of the appendicular skeleton include a scapulocoracoid and cleithrum, humerus, pelvis and femur. The inferred relation of the cleithrum to the clavicle is not of the normal labyrinthodont type. The species has a characteristic massive dermal armour. Two dermal shields are preserved, together with a large number of scutes, some singly and some sutured as transverse bands. The armour is reconstructed as pectoral and pelvic shields connected by neural and costal bands. A reconstruction of the whole skeleton is attempted and the possible mode of life, that of a heavy terrestrial form, is discussed. A relation between Peltobatrachus and the Triassic plagiosaurs is postulated on both cranial and postcranial characters. The most significant common features are considered to be the pustular ornament, the ossification of the endochondral bones of the occiput, the form of the vertebrae and that of the dermal armour. Establishment of this relationship removes the plagiosaurs from the stereospondylous labyrinthodonts in agreement with Nilsson (1946). The taxonomic position of the plagiosaurs is discussed and a case made against their relationship to the brachyopids. It is also concluded that they are not closely related to Dvinosaurus or to the metoposaurs. The position of the genera Plagiosternum and Taphrognathus is discussed in the light of the diagnostic features of the group. The retention of the plagiosaurs (including Peltobatrachus) within the Labyrinthodontia as defined by Romer is proposed, but their separation from temnospondyls and anthracosaurs as the order Plagiosauria is endorsed. A new diagnosis of the order is given and diagnostic characters of two constituent suborders, the Peltobatrachi (including only Peltobatrachus) and the Plagiosauri are listed.