A study has been made of the formation and regression of synapses formed by spinal nerves 16 and 17 in axolotl hind limb flexor muscles following the severing of nerve 16, using histological, ultrastructural and electrophysiological techniques. Axolotl hind limb flexor myofibres possessed 'en plaque' end plates from either spinal nerve 16 or 17 or both at intervals of about 1000 μm along their length; the myofibre's length constant was about 700 μm allowing electrophysiological observations of at least 2 of these synapses during a single impalement; transmitter release at these synapses could be described by binomial statistics and in a given set of ionic conditions the binomial statistic parameter n was directly proportional to the size of the nerve terminals whilst the binomial statistic parameter p was invariant to changes in nerve terminal size. The distribution of synapses formed by spinal nerves 16 and 17 in different sectors of the axolotl hind limb flexor muscles were determined from a study of evoked end plate potentials; the middle and proximal sectors of the flexor muscles contained myofibres which received an innervation from nerve 16 only, whereas the sectors surrounding these contained myofibres innervated either by nerve 16 or nerve 17 or by both nerves. 6 Days following the severing of spinal nerve 16, evoked transmitter release from the synapses formed by this nerve had failed; transmission was subsequently recorded at a few synapses formed by nerve 17 in the middle and proximal sectors of the flexor muscles which are not normally innervated by this nerve and these synapses had a low n; during the succeeding 4 wk the value of n at the synapses increased to a size about 70% that of the terminals normally formed by nerve 16 at these sites. 4 Wk after severing nerve 16, myofibres which possessed synapses formed by nerve 17 also possessed synapses from reinnervating nerve 16 and these were sometimes formed at the same synaptic sites as those occupied by nerve 17. In the subsequent 16 wk, the n value of synapses formed by nerve 17 declined whilst the n values of synapses formed by innervating nerve 16 on the same myofibres matured to their control size. It is suggested that on severing nerve 16 collateral sprouting of nearby intact nerve 17 occurs and these collateral sprouts innervate the denervated synaptic sites, although the sprouts are not as well matched to the denervated synaptic sites as are the original nerve terminals; thus if nerve 16 returns it preferentially forms synapses at its original synaptic sites, and the collateral synapses formed by nerve 17 regress.