Groups and sub-groups (clusters) of small granule-containing cells (“small cells”) were analysed at 3 and 6μm intervals and in serial sections, in rats aged 2–13 months. Fully intraganglionic clusters of small cells were all found to receive an incoming (“afferent”) innervation, of the order of 3–6 afferent terminals per cell, derived from axons of preganglionic type via multifocal. symmetrical, mainly axosomatic synapses. No evidence was obtained of sharing of preganglionic inputs between small cells and principal neurones. Intraganglionic clusters also regularly gave outgoing (“efferent”) synapses of the asymmetrical type, of the order of 2–6 per cell, to intraganglionic nerve elements; 30–50% of these synapses were given from somata, 50–70% from processes of the small cells. Whenever the postsynaptic structure was identifiable these synapses were all found to be given to postganglionic neurones or their dendrites, principally to spine-like processes or slender twigs. In some ganglia a few efferent synapses to other small cells were observed; these were of the symmetrical type. Efferent synapses to nerve profiles resembling chemosensory axon terminals, also of the symmetrical type, were extremely infrequent (fewer than 1% of all efferent synapses) in intraganglionic small cell groups and appeared virtually restricted to glomus-like clusters of small cell, which lay intracapsularly, or in and near the bases of nerves entering or leaving the ganglion. Almost all groups and clusters of small cells were located near to fenestrated capillary vessels, which are not found elsewhere in the ganglion. The implications of possible non-synaptic release of material from small cells via membrane regions not covered by satellite cell cytoplasm, were explored in a nearest-neighbour analysis. These “exposed” regions comprised 1–3% of the small cell surface, a proportion comparable with those engaged in receiving afferent synapses or in giving efferent synapses. The majority of such regions faced toward other nerve profiles (axons and dendrites) ensheathed in satellite cytoplasm (mean, 30%), intraganglionic tissue spaces wider than 3 μm (mean, 30%) or other small cells (mean, 14%); 25% faced toward blood vessels, but of these vascularly directed regions, only one fifth (or 5% of the total) on average faced directly toward fenestrated endothelium, the rest being non-fenestrated and/or separated by pericyte processes from the exposed regions of small cell membrane. Thirty-three percent of the small cells in a sample of 242 lay within 2μm of the nearest blood vessel. Thus the vascular relationship is close, but vascular fenestrations would not appear to provide the major or most direct route for the distribution of secretions from the small cells. Serial section analyses showed variation between small cells within a cluster and also between clusters in the extent of their synaptic and other relationships: some small cells lacked exposed membrane, some gave no efferent synapses, whereas others exhibited both features simultaneously. In older rats (up to 13 months) the afferent synaptic terminals upon the small cells were larger and gave more synaptic foci, though the terminals themselves were not more numerous; points of efferent synapse from the small cells increased in incidence; processes of the small cells increased in volume and incidence; and there was evidence of increasing separation of small cells, with a shift of intercellular attachments from the somata out on to the processes. The proportion and distribution of regions of exposed membrane remained relatively constant. The cytoplasmic volume of the small cells was found to vary according to the incidence of afferent synaptic terminals and to the weight of the rat. These findings are discussed in relation to the possible actions of the small cells. It is concluded that the small granule-containing cells in this ganglion of the rat are consistently both interneurone- and paraneurone-like, exerting a focused synaptic influence upon the principal neurones in addition to a more diffuse non-synaptic influence which may include both preganglionic axons and postganglionic neurones, as well as other small cells and the ganglionic vasculature. Physiological and pharmacological evidence suggests that their effects on ganglionic transmission may be directly inhibitory or indirectly facilitatory, or both; the present observations suggest that these effects can be precisely recruited, possibly differentially from principal neurone activation, and may be directed to selected pathways through the ganglion.