Acid-sensing ion channels (ASICs) are trimeric cation-selective channels activated by extracellular acidification. Amongst many pathological roles, ASICs are an important mediator of ischemic cell death and hence an attractive drug target for stroke treatment as well as other conditions. A peptide called Hi1a, isolated from Australian funnel web spider venom, inhibits ASIC1a and attenuates cell death in a stroke model up to 8 h after stroke induction. Here, we set out to understand the molecular basis for Hi1a's action. Hi1a is a bivalent toxin with two inhibitory cystine knot domains joined by a short linker. We found that both Hi1a domains modulate human ASIC1a gating with the N-terminal domain impairing channel activation while the C-terminal domain produces a "pro-open" phenotype even at submicromolar concentrations. Interestingly, both domains bind at the same site since a single point mutation, F352A, abolishes functional effects and reduces toxin affinity in surface plasmon resonance measurements. Therefore, the action of Hi1a at ASIC1a appears to arise through a mutually exclusive binding model where either the N or C domain of a single Hi1a binds one ASIC1a subunit. An ASIC1a trimer may bind several inhibitory N domains and one or more pro-open C domains at any one time, accounting for the incomplete inhibition of wild type Hi1a. We also found that the functional differences between these two domains are partially transferred by mutagenesis, affording new insight into the channel function and possible novel avenues of drug design.