Glutamatergic synapses are of immense importance in that they mediate most excitation in the vertebrate brain. They are remarkably diverse. Some produce large, precisely timed synaptic currents, while others produce small, slow currents. Some are indefatigable, signalling robustly and with high fidelity, independent of how rapidly they are driven while others are plastic, constantly strengthening or weakening as a function of whether they are driven rapidly or slowly. α-Amino-3-hydroxy-5-methyl-4-isoxazole proprionate (AMPA) receptors report the presynaptic release of neurotransmitter to postsynaptic neurons. They are tetrameric ion channels formed as combinations of four α subunits, GluA1–4, each of which has the alternatively spliced forms, flip and flop. The subunits differ in the length and in the sequence of their cytoplasmic, carboxy terminal tails which mediate interactions with other molecular components of synapses that regulate receptor trafficking and tethering (Jackson & Nicoll, 2011). The receptor subunit composition determines the speed of synaptic signalling; its receptor number determines the strength of a synapse. It is essential for some functions of the brain that glutamatergic synapses not be altered by synaptic traffic. At the early stages of the auditory pathway, for example, acoustic information must be transmitted faithfully, without distortion by previous sounds, for that information to be useful for computing the location and meaning of sounds. Acoustic information from the cochlea is relayed through the cochlear nucleus and the contralateral medial nucleus of the trapezoid body (MNTB) to the contralateral lateral superior olive. The lateral superior olive compares the intensity of sounds at the two ears for localizing sound sources in the horizontal plane. MNTB neurons are excited through a single sumptuous, calyceal synapse that contacts the smooth surface of the target soma, the calyx of Held (Fig. 1A). This synapse has become a darling of synaptic physiologists because the presynaptic terminals are accessible to electrodes but has been largely ignored by cell biologists because it is boringly stable and not experimentally tractable. Figure 1 Glutamatergic, excitatory synapses are diverse As at upstream synapses, AMPA receptors in mature MNTB neurons are formed from GluA3 flop and/or GluA4 flop subunits in which GluA1 and GluA2 are not functionally evident (Joshi et al. 2004; Koike-Tani et al. 2005). In this issue of The Journal of Physiology, Yang et al. (2011) compare EPSCs in the MNTB of wild type mice with those in mice lacking GluA3 (GluA3 KO) and in mice lacking GluA4 (GluA4 KO). They find that in the MNTB of GluA4 KO mice, EPSCs are dramatically diminished in size and slowed whereas in GluA3 KO mice, EPSCs are only subtly slowed. GluA2 substituted to a small extent for the loss of GluA3 but not detectably for the loss of GluA4. The calyceal synapse does not rely entirely on GluA4, however, because the removal of GluA3 results in greater synaptic depression and desensitization than in the wild type. Subtle but detectable differences in presynaptic functions in KO mice serve as a reminder that differences between wild type and GluA4 KO mice reflect not only the removal of GluA4 at this synapse but can also reflect changes upstream in the auditory pathway as well as compensatory changes at all levels. These findings leave no doubt, however, that GluA4 is essential in mature, excitatory synapses in the MNTB, that GluA3 plays a minor role, and that GluA1 and 2 are almost undetectable in the MNTB. Cell biologists have been attracted by the dynamic glutamatergic synapses on spines in hippocampal pyramidal cells (Fig. 1B). They have studied the trafficking of AMPA receptors in hippocampal synapses where endogenous AMPA receptors are formed from either GluA1/2 or GluA2/3 subunits. Spines give neighbouring synapses biochemical privacy. The long, cytoplasmic, carboxy terminal tails of GluA1 subunits interact with proteins that drive AMPA receptors into synapses as a function of activity (Hayashi et al. 2000). GluA2/3 AMPA receptors cycle constitutively, replacing GluA1/2 receptors. The short cytoplasmic tails of GluA2 subunits regulate the interaction with a different group of synaptic proteins that promote the removal of GluA2-containing AMPA receptors (Kim et al. 2001). The relative rates of insertion and excision of receptors determines how many AMPA receptors are at a synapse and how much current it delivers at any moment. The findings of Yang et al. show that calyceal synapses of MNTB neurons manage without either GluA1 or GluA2, which are so important in the hippocampus, and do pretty well in the absence of GluA3. At these synapses fleets of the speedy GluA4 flop subunits seem to be coordinated by synaptic proteins in the communal postsynaptic environment of the soma.