Fast synaptic transmission in the brain is mediated mainly by AMPA receptors (AMPARs). There is compelling evidence that neuronal AMPARs are usually associated with transmembrane AMPA receptor regulatory proteins (TARPs), which control their trafficking, as well as their biophysical and pharmacological properties (Nicoll et al., 2006). To understand some of the mechanisms that underlie the activation of TARP-associated AMPARs, we have examined the behaviour of single-channels recorded from homomeric GluA1 AMPARs co-expressed with TARP family members γ-2 (stargazin), γ-4, and γ-5. GluA1/TARP single-channel currents were examined in outside-out patches from tsA201 cells transfected with cDNA encoding green fluorescent protein, GluA1 and either γ-2, γ-4, or γ-5. Receptors were activated with saturating (10 mM) glutamate. Single-channel individual apparent open times were 4.0 ± 0.6 ms (GluA1/γ-2, n = 4), 6.2 ± 2.8 ms (GluA1/γ-4, n = 7), and 2.7 ± 1.0 (GluA1/γ-5, n = 3). The mean duration of bursts for GluA1/γ-2 was 6.4 ± 1.6 ms (n = 4) and 5.0 ± 2.7 ms for GluA1/γ-5 (n = 3). γ-4-containing AMPARs produced much longer bursts, some of which lasted hundreds of milliseconds, with an overall mean burst length of 19.4 ± 3.5 ms (n = 7). All of the TARP-associated GluA1 receptors we have examined exhibited multiple single-channel conductance levels. These ranged from 7 to 51 pS, with weighted mean conductances of 23.0 ± 6.5 pS (γ-2, n= 4), 24.8 ± 1.7 pS (γ-4, n = 7), and 21.8 ± 4.1 pS (γ-5, n = 3). We found that GluA1/γ-4 receptors exhibited sub-conductance levels of exceptionally long duration. In addition, our single-channel recordings showed the presence of ‘mode changing’ under steady-state conditions, in which channels spent extended periods of time at one (mean) conductance level before switching to a higher, or lower, conductance level. The long duration sub-conductance opening associated with GluA1/γ-4 receptors are difficult to explain in terms of previous models of AMPAR channel gating, in which the conductance level is directly related to agonist occupancy (Rosenmund et al., 1998; Smith & Howe 2000; Gebhardt & Cull-Candy, 2006). Instead our results are compatible with the hypothesis that conductance level is correlated with ligand-binding domain closure (Zhang et al., 2008). We have analyzed the frequency of transitions to construct ‘conductance-dependent’ kinetic models of the different GluA1/TARP receptors and their modal gating.