The LGN is not a simple, machine-like relay of retinal information to cortex: it instead performs a dynamic, variable relay that presumably reflects the varying behavioral needs of the visual system. Two aspects of this will be reviewed. First is the appreciation that LGN (and other thalamic) relay cells have a variety of voltage-dependent membrane conductances that dramatically alter the way they respond to retinal inputs. Perhaps the most important next to the those underlying the action potential is a voltage-dependent Ca2+ conductance. This can be activated by depolarization (e.g., from a retinal EPSP) only from a hyperpolarized level (e.g., more negative than about -65mV), because the Ca2+ conductance becomes inactive at more depolarized levels. The activation state of this conductance determines whether the relay cell responds in tonic or burst firing mode to its retinal inputs, and which response mode is operating is of critical importance to the nature of information relayed to cortex. Second is the fact that, while the information relayed through LGN to cortex is retinal in origin, retinal synapses represent only about 7% of inputs to relay cells. The vast majority of inputs derive from intrinsic local GABAergic cells, from visual cortex, and from the brainstem, and they each contribute about 30% of all synapses on relay cells. These nonretinal inputs are modulatory in nature and, among other things, serve to control membrane potential, thereby controlling voltage dependent conductances like the Ca2+ conductance described above. An attempt will be made to bring these observations together in a testable hypothesis that addresses how and under what conditions nonretinal inputs influence the relay through the LGN.