The damaged septohippocampal pathway was utilized to study the axonal regeneration of injured neurons. Semipermeable tubes, 2-mm long, were placed in the axis of the transected septohippocampal pathway of adult rats. In a first series of experiments, empty tubes were implanted. Even six weeks after the operation, no regenerated axons were observed in the conduit. In a second series of experiments, in order to validate our approach, segments of pre-degenerated sciatic nerves were introduced into the tubes. Under these experimental conditions, acetylcholinesterase (AChE)-containing regenerated axonal processes were detected in the grafted sciatic nerves. Glial fibrillary acidic protein (GFAP)-immunodetection showed that astroglial cells and astrocyte processes were able to progress on and into the peripheral grafts. At the electron microscopic level, axons were observed in close contact with Schwann cells which myelinated some of them. In some other cases, unmyelinated axons were also present at the surface of reactive astroglial cells filled by numerous intermediate filaments. These central glial cells had migrated among the sciatic nerve collagen fibers. No axon was detected without glial cell contact. In a third series of experiments, we implanted semipermeable tubes previously filled with a fibrin-fibronectin-containing matrix provided by peripheral regeneration chambers. One week after the implantation of the tubes containing this peripheral substrate, different cell types were observed migrating into the conduit and replacing the fibrin-fibronectin-containing matrix. Among these cells astrocytes were present as revealed by GFAP-immunocytochemistry and electron microscopic examinations. During the following weeks, axons were detected in contact with the reactive astroglial cells. AChE-histochemistry showed that axons were able to cross the two millimeter distance separating the septal part and the hippocampal part of the lesion site. GABA (γ-aminobutyric acid)-ergic fibers were also detected in the regenerated structure. These experiments show that cellular or acellular substrates provided by the PNS can promote the regeneration of CNS GABAergic and cholinergic neurons. Our observations suggest that astrocytes can take an important part, after their migration or after extending processes, in the axonal regeneration in the adult CNS of the rat, possibly in furnishing a cellular terrain for the progression of growth cones over a distance of two millimeters and in maintaining regenerated axons at least until the sixth week after the operation.