The effects of interrupting the axons of principal neurones in the superior cervical ganglion of adult guinea-pigs were studied by means of intracellular recording, and light and electron microscopy. 1. Within 72 hr of axon interruption, the amplitude of exitatory postsynaptic potentials potentials (e.p.s.p.s) recorded in principal neurons in response to maximal preganglionic stimulation declined. E.p.s.p.s were maximally reduced (by more than 70% on average) 4-7 days following interruption, and failed to bring many cells to threshold. E.p.s.p.s. recorded in nearby neurones whose axons remained intact were unaffected. 2. In ganglia in which axon interruption was achieved by means of nerve crush (thus allowing prompt regeneration), mean e.p.s.p. amplitudes began to increase again after about 1-2 weeks. One month after the initial injury many neurones had e.p.s.p.s of normal amplitude, and by 2 months affected neurones were indistinguishable from control cells. Functional peripheral connexions were re-established during the period of synaptic recovery. 3. The mean number of synapses identified electron microscopically in ganglia in which all the major efferent branches had been crushed decreased by 65-70% in parallel with synaptic depression measured by intracellular recording. However synapse counts did not return to normal levels even after 3 months. 4. During the period of maximum synaptic depression, numerous abnormal profiles which contained accumulations of vesicular and tubular organelles, vesicles, and mitochondria were observed in electron microscopic sections. Injection of horseradish peroxidase into affected neurones demonstrated dendritic swelling which probably correspond to these profiles. 5. Little or no difference was found in the electrical properties of normal neurones and neurones whose axons had been interrupted 4-7 days previously. However, the mean amplitude of spontaneously occurring synaptic potentials was reduced, and the amplitude distribution was shifted. This abnormality of the synapses which remain on affected neurones also contributes to synaptic depression. 6. Counts of neurones in normal and experimental ganglia showed that approximately half the principal cells died 1-5 weeks after crushing the major efferent brances. This finding presumably explains the failure of synapse counts to return to control levels after recovery. 7. If axons were prevented from growing back to their target organ by chronic ligation, surviving neurones whose axons were enclosed by the ligature did not generally recover normal synaptic function. Following ligation, most affected cells died within a month. 8. Thus the integrity of a principal cell's axon is necessary for the maintenance of preganglionic synaptic contacts, and ultimately for neuronal survival. The basis of neuronal recovery from the effects of axon interruption appears to be some aspect of regeneration to the peripheral target.