Calretinin immunoreactivity is almost completely confined to two classes of neuron in the myenteric plexus of the guinea-pig small intestine, longitudinal muscle motor neurons and ascending interneurons. Nerve cell bodies of the two classes can be readily identified by their sizes and positions in ganglia. The motor neurons, which are small Dogiel type I neurons, are about 20% and the interneurons, which are medium-sized Dogiel type I neurons, are about 5% of myenteric neurons. In the present work, we have also discovered a minor population (0.1%) of small filamentous neurons. In unoperated regions of intestine, at the light microscopic level, numerous calretinin immunoreactive nerve fibres were found in the tertiary plexus that innervates the longitudinal muscle and a medium density of varicose fibres formed pericellular endings in the myenteric ganglia. After double myotomy operations, in areas of plexus 0.5 to 1.5 mm wide which were isolated from ascending and descending inputs, calretinin-immunoreactive fibres of the tertiary plexus were unchanged, but the pericellular endings in the ganglia disappeared. Both the ascending interneurons and the longitudinal muscle motor neurons received ultrastructurally identified synapses and close axonal contacts that were calretinin-immunoreactive. These were counted in semi-serial sections from normal intestine and from regions between myotomy operations. In unoperated intestine, the proportions of calretinin-immunoreactive synapses on small, calretinin-immunoreactive, Dogiel type I nerve cells and small filamentous nerve cells were 30% and 0.1% respectively and on medium-sized Dogiel type I cells the proportion was 28%. Electron microscopy revealed an almost complete loss of immunoreactive inputs to the small Dogiel type I cells between double myotomies, but the number of unreactive inputs was the same as in normal intestine. This work demonstrates that the ascending calretinin-immunoreactive interneurons connect with one another to form ascending chains in the myenteric plexus and that they also provide about 1/3 of the inputs received by calretinin-immunoreactive longitudinal muscle motor neurons. Many of the remaining inputs to these motor neurons are local; we have deduced that these are mainly from primary sensory neurons.