In an ultrastructural immunocytochemical study we investigated the development of the gonadotropic cells in the pituitary of two to six months old male African catfish in relation to testicular development. In this period, pituitary and testicular tissue samples were collected on five occasions (groups I-V). Blood samples could only be taken from the fish in groups III-V. The testicular development was divided in three stages i.e., immature (only spermatogonia, group I), early (spermatogonia and spermatocytes, groups II and III) and advanced (all germ cell stages including spermatozoa, groups IV and V) spermatogenesis. 11-Ketotestosterone blood levels were low, except for the last group. Antisera were raised against the complete catfish α,βGTH-II, as well as to the separate α- and β-subunits of catfish GTH-II. In the proximal pars distalis of immature fish, undifferentiated cells, somatotrops, putative thyrotrops (pTSH) and putative gonadotrops (pGTH) were found. In the two latter, secretory granules were labeled with anti-αGTH, but not with anti-βGTH-II. pTSH- and pGTH-cells were distinguished on the basis of the size of their secretory granules. During early spermatogenesis, two classes of putative gonadotrops could be distinguished. One type had the same immunocytochemical and ultrastructural characteristics as in immature fish; the secretory granules in the second cell type, which was more abundant, were also immunopositive for anti-βGTH-II. The mean volume of the secretory granules in these GTH-II cells was three times larger than that in the early appearing pGTH-cells. In addition, the later appearing GTH-II cells contained large inclusions, known as globules. These structures labeled with anti-αβGTH-II and with anti-βGTH-II, but not with anti-αGTH. It is assumed that the globules are involved in a differential storage and/or breakdown of the GTH-II subunits. During advanced spermatogenesis the two gonadotropic cell types could still be distinguished, but the early appearing pGTH-cell type was scarce. The present observations permit the conclusion that the early appearing cells may be GTH-I cells. However, definitive proof about their identity depends on the availability of antibodies or cDNA probes specific for GTH-I.