Sensory information is processed and encoded by neural networks. In order to understand how the nervous system is able to rapidly integrate and store sensory information, knowledge of the connections and properties of the neurons in these circuits is required. The fruit fly Drosophila melanogaster provides a particularly powerful species to investigate the neural circuits of the olfactory system because in addition to possessing a simple olfactory system amenable to circuit analysis, a host of genetic reagents are available, including the GAL4-UAS system for targeted gene expression. The mushroom bodies, paired structures historically implicated in olfactory learning and memory, receive olfactory information at the mushroom body calyx from second-order olfactory projection neurons (PNs). Within the calyx, individual PN axonal boutons are surrounded by dendritic arborizations from intrinsic Kenyon cells (KCs) and each tiny cluster constitutes a single microglomerulus. Cells that connect the calyx with other areas of the brain, extrinsic neurons (ENs), also contribute to microglomeruli. Most of these contain the neurotransmitter, GABA, and are presumed to be inhibitory. In this study, the synaptic characteristics, neural circuits, and plasticity of calycal cells have been investigated using a combination of serial section electron and confocal microscopy. The findings reveal several new features of the circuits in the calyx: 1) The calyx contains three ultrastructurally distinct types of PN boutons that are heterogeneous in shape and exhibit subtle differences in synaptic densities. 2) All PN boutons form both ribbon and non-ribbon synapses, and from their smaller size and fewer postsynaptic partners, non-ribbon synapses may possibly become converted to ribbon synapses after activity; the olfactory signal may then be transmitted more strongly and efficiently at ribbon synapses. 3) PN boutons with an electron-dense cytoplasm have the most ribbon synapses per unit area of membrane as well as the highest ratio of ribbon to non-ribbon synapses, and thus may be more active and efficient than other boutons. 4) KC neurites are not exclusively postsynaptic in the calyx and can form occasional ribbon synapses, the functional interpretation of which awaits identification of their postsynaptic partners and vesicle contents. 5) Each PN bouton may contribute input to a single dendritic KC claw at about three presynaptic sites. For the postsynaptic side, a single claw receives input from individual presynaptic sites that must be highly redundant. 6) There may be important processing of the olfactory signal by local circuits formed by ENs in the calyx; ENs form synaptic connections with PNs, KCs, and other ENs. 7) Extensive serial synapses link EN terminals into a network, presumed to be GABAergic and inhibitory, that extends between microglomeruli and may be autaptic. 8) The structure and synaptic connectivity of microglomeruli may undergo changes after adult emergence. 9) vGAT and GAD1-GAL4 lines drive ectopic expression of marker genes in KCs and are not reliable reporters of GABA-positive cells. 10) Previously identified calycal ENs (MB-C1, MB-C2/C3, MB-CP1) are not immunopositive for GAD1, a marker of GABA-containing cells. 11) A network of ENs expressing a GABA phenotype differently innervates anatomically and functionally discrete areas of the honeybee calyx, and in addition the density of innervation may change with alterations in age and/or experience.