A pituitary cell population of 14-day-old female rats, containing lactotropes and somatotropes but deprived of gonadotropes, was prepared by unit gravity sedimentation through a serum albumin gradient and allowed to reaggregate either as such or after mixing this population with cells of the gonadotropic αT3-1 cell line. In a perifusion system gonadotropin-releasing hormone (GnRH) had no effect on prolactin (PRL) and growth hormone (GH) release in the former aggregates but stimulated PRL release in the latter. In the αT3-l cell-containing aggregates GnRH showed a biphasic effect on GH release: inhibition during exposure to GnRH followed by a rebound secretion upon removal of the peptide. The aggregation capacity of αT3-l cells with the normal pituitary cells was demonstrated by using an αT3-l cell clone stably transfected with the reporter gene β-galactosidase. Perifusion of the gonado-trope-deprived aggregates with medium conditioned by αT3-1 cell provoked a rapid stimulation of PRL release and a biphasic effect on GH release. Medium conditioned by the cortico-tropic cell line AtT20 also stimulated PRL release but had no concomitant effect on GH release. Medium conditioned by αT3-l cells, when added for 40 h to aggregates of 14-day-old rat pituitary, provoked an increase in the number of <sup>3</sup>H-thymidine (<sup>3</sup>H-T)-labelled lactotropes and a decreased in the number of <sup>3</sup>H-T-labelled somatotropes. The conditioned medium was concentrated on Sep-Pak C18 and ultrafiltrated through an Amicon membrane with 3-kD molecular weight cut-off and the retained molecules separated by reversed-phase HPLC. The material stimulating <sup>3</sup>H-T labelling of lactotropes eluted from the column with a different retention time than material inhibiting <sup>3</sup>H-T labelling of somatotropes, suggesting that the effect on lactotropes is mediated by (a) molecule(s) different from that affecting somatotropes. The effects of αT3-l cells and their secretion products on lactotropes and somatotropes were comparable to those we previously observed using enriched populations of normal gonadotropes. The HPLC elution profiles of the substances affecting <sup>3</sup>H-T incorporation as well as the specificity of these effects were also similar to that of the substances isolated previously from gonadotrope-conditioned medium. The present data, therefore, support previous conclusions on the paracrine control of the lactotrope/somatotrope lineage by the gonadotropes. Further purification and chemical characterization of the growth factors with selective action on lactotropes and somatotropes may lead to a better understanding of the development of the latter lineage.