1. Intact normal innervation of muscle fibres and other peripheral targets usually prevents regenerating nerves from forming synapses with the targets. Whether intact innervation similarly prevents synapse formation on central target neurones has rarely been tested. This question was examined here for interneurone A of the crayfish last abdominal ganglion. 2. Interneurone A normally receives synaptic input from mechanoreceptor neurones distributed over the side of the tailfan ipsilateral to interneurone A's axon and unilateral dendrites. When the five nerve roots carrying mechanoreceptor axons of one side are cut and central and peripheral ends of one or more are sutured together, regeneration and reinnervation of interneurone A occurs over some two to six weeks. If peripheral ends of roots from the 'wrong' (contralateral) side of the body are sutured to ipsilateral central stumps, they also form connexions with interneurone A. When roots from the two sides of the body are simultaneously tied to a central stump, functional connexion formation occurs equally well for afferents from both sides. Therefore, roots of the two sides seem to be equivalent in their ability to reinnervate interneurone A. 3. If peripheral ends of roots from one side of the tailfan are tied to roots on the intact opposite side of the body, the cut axons appear to grow into the last ganglion but usually do not form functional synapses there. The intact innervation therefore seems to exclude further innervation by other acceptable afferents. 4. It is known that mechanoreceptors are added to the tailfan at moult. Exclusion of extra innervation often broke down partially in animals that moulted during the present experiments. This suggests the possibility that synapse formation or exchange may be controlled by moult-inducing hormones.