GABAA receptors (GABAARs) are the major inhibitory neurotransmitter receptors in the brain and are therapeutic targets for many indications including sedation, anesthesia and anxiolysis. There is, however, considerable scope for the development of new therapeutics with improved beneficial effects and reduced side-effect profiles. The anthelminthic drug, ivermectin, activates the GABAAR although its binding site is not known. The molecular site of action of ivermectin has, however, been defined by crystallography in the homologous glutamate-gated chloride channel. Resolving the molecular mechanisms of ivermectin binding to α1β2γ2L GABAARs may provide insights into the design of improved therapeutics. Given that ivermectin binds to subunit interfaces, we sought to define (1) which subunit interface sites it binds to, (2) whether these sites are equivalent in terms of ivermectin sensitivity or efficacy, and (3) how many must be occupied for maximal efficacy. Our approach involved precluding ivermectin from binding to particular interfaces by introducing bulky M3 domain 36′F sidechains to the “+” side of those interfaces. We thereby demonstrated that ivermectin produces irreversible channel activation only when it binds to the single γ2L-β2 interface site. When it binds to α1-β2 sites it elicits potentiation of GABA-gated currents but has no irreversible activating effect. Ivermectin cannot bind to the β2-α1 interface site due to its endogenous bulky 36′ methionine. Replacing this with an alanine creates a functional site at this interface, but surprisingly it is inhibitory. Molecular docking simulations reveal that the γ2L-β2 interface forms more contacts with ivermectin than the other interfaces, possibly explaining why ivermectin appears to bind irreversibly at this interface. This study demonstrates unexpectedly stark pharmacological differences among GABAAR ivermectin binding sites.