The development of the jugular and cerebral veins and adjacent lymphatic vessels in Hepsetus, Salmo, Amia, Acipenser, Neoceratodus, Scyllium and Squalus has been studied on a large embryological material. The reconstruction method has been applied. There has been much confusion in the literature on head veins. This is partly due to the rather schematic investigations of e.g. van Gelderen, and partly to the publications of de Beer and Holmgren, paying no regard to the lymphatic vessels. The currently new features on the v. capitis medialis are that in adult fishes a larger part of the vein remains than was earlier believed. It usually consists of three portions. No secondary vein takes part in the constitution of the jugular vein, as was supposed by de Beer, Holmgren and others. In actinopterygians the secondary vein is in reality represented by the anterior portion of the v. cardinalis anterior as well as of all the v. capitis lateralis, and in elasmobranchs it is only represented by the former portion. The elasmobranchian (except Squalus) jugular vein is instead formed from the v. capitis lateralis, and in teleostomes and Squalus it is developed from two veins, viz. the anterior portion of the v. capitis medialis and all the v. capitis lateralis. The latter vein is formed from two (teleostomes) or three (elasmobranchs) loops. The v. cardinalis anterior becomes completely atrophied in Squalus, and partly atrophied in Acipenser, Amia and Neoceratodus. The development of the three main cerebral veins is described. van Gelderen's classification of the v. cerebralis media is criticized. And so also is his hypothesis for the evolution of the head veins in fishes. It is suggested that the Hepsetus type is primitive and that the Acipenser and Amia types have evolved from it or from a similar type. Neoceratodus and Squalus have rather specialised types, which seem to have evolved separately right from the base of the system.