The mammalian olfactory bulb (OB) generates gamma (40 – 100 Hz) and beta (15 – 30 Hz) oscillations of the local field potential (LFP). Beta oscillations occur in response to odorants in learning or odor sensitization paradigms, but their generation mechanism is still poorly understood. When centrifugal inputs to the OB are blocked, beta oscillations disappear, but gamma oscillations persist. These inputs primarily target GABAergic granule cells (GC) in the GC layer (GCL) and regulate their excitability. This leads us to the central question motivating this work: What role does GC excitability play in generating beta oscillations? To answer this question we first developed a computational model incorporating the biophysical properties of the reciprocal dendrodendritic synapses between glutamatergic mitral cells (MC) and GCs. The model predicted that beta oscillations emerge only when inhibition of MCs, due to heightened GC excitability, and the excitation of MCs due to sensory input is sufficiently balanced. Because of this balance of excitation and inhibition, the model predicted that beta oscillations could also be supported in the absence of heightened GC excitability provided that the input strength was also low. The model also predicted that beta oscillations are sustained by voltage dependent calcium channel (VDCC) mediated GABA release, independently of NMDA channels.,We tested the predictions of this model using pharmacology in the OBs of rats. Infusion of scopolamine, a muscarinic antagonist known to decrease GC excitability, decreased or completely suppressed odor-evoked beta in response to a strong stimulus, but increased beta power in response to a weak stimulus, as predicted by our model. APV, an NMDA receptor antagonist, suppressed gamma oscillations selectively (in OB and PC), lending support to the model’s prediction that beta oscillations can be supported by VDCC currents.,In another set of experiments we recorded extracellular potentials in the GCL of rats using multichannel Si probes. Because these were the first recordings of GCs and other GCL interneurons in awake, freely behaving rats, the nature of these experiments was exploratory. We found that many GCL neurons fired at the onset of beta oscillations, which is consistent with our model, because GC excitability increases can be triggered by GC somatic spikes. We also found a rich diversity of excitatory and inhibitory responses that showed odor selectivity and phase locking to different LFP frequency bands. Some responses evolved over the course of multiple days. By classifying responses to different odors using a distance metric analysis, we showed that some cells were better at distinguishing between odors based on their firing rates, while others were better based on spike timing. Intriguingly, the timescales at which most of these spike-timing cells best distinguished between odors were in the theta and beta frequency ranges. In a final set of experiments we trained two rats implanted with silicon (Si) probes to poke their noses into an odor port to receive sugar pellet rewards. We found that some GCL neurons fired precisely at the time when a rat poked its nose into the odor port, while others fired only at the onset of beta oscillations. In one rat we also found cells that appeared to change their response when reward was discontinued. Together, these experiments lend strong support for the main predictions of the model, and also provide exciting preliminary data for future studies regarding the involvement of GCL neurons in contextual representations of odors during motivated behaviors.