At many excitatory synapses in the CNS, glutamate activates both AMPA- and NMDARs. Although it is commonly assumed that these receptors are co-localized in the postsynaptic membrane, the situation at adult synapses is unclear. Our studies of cerebellar granule cells (GCs) show that in immature animals NMDARs are activated during both spontaneous mEPSCs and following mossy fibre (MF) stimulation. By contrast, in mature mice, MF activity leads to activation of NMDARs (predominantly NR2C-containing; Cathala et al. 2000) but mEPSCs lack an NMDAR-mediated component (Cathala et al. 2003). Here we re-examined mEPSCs and evoked EPSCs in GCs from young and mature mice, and undertook EM immunogold localization of NMDARs, to investigate the presence and localization of NMDARs in GCs of adult mice. Patch-clamp recordings were made from GCs in cerebellar slices from immature (P7-9) or mature (P34-55) mice, at 35C. While mEPSCs were readily detected in Mg2+ free conditions at -70mV at both ages, these events were abolished in mature GCs by CNQX (non-NMDAR blocker), suggesting that at this age GCs lack synaptic NMDARs. To determine if NMDARs are present in the synaptic membrane, but have their activity suppressed, we examined mEPSCs under conditions that increase the likelihood of activating and detecting NMDAR currents. We examined cells in extracellular Mg2+ (at +30mV) to enhance NMDAR activation, with increased glycine to maximize activation of the NMDAR glycine site, in conditions that suppress tonic H+ inhibition of NMDARs and in conditions that suppress mGluR inhibition of NMDARs. No NMDAR-mediated component was seen in any of these conditions. This was also the case when Sr2+ was used to desynchronize release and enable examination of MF-evoked quantal events, suggesting that this discrepancy does not reflect a difference between spontaneous and evoked release. Mature GCs are thought to express mRNA for both NR2C and NR2A. Consistent with this, we found both high- (50pS) and low conductance (19 and 38pS) NMDAR channels in outside-out patches from these cells. Immunogold labeling confirmed the presence of multiple NMDAR subunit types in the extrasynaptic membrane, and revealed a high particle density at intraglomerular non-synaptic attachment plaques (between adjacent dendrites). NR1, NR2A and NR2C labeling was also detected within postsynaptic densities (PSD), but at much lower level. Overall our data suggest that either the sparse labeling of PSD does not reflect the presence of functional receptors, or the level of synaptic NMDAR expression is below that which can be resolved during our recordings. Alternatively, NMDARs are present but not activated during the brief glutamate transient arising from the release of single quanta. Future modeling studies will address this question.