Twenty-five avermectin analogs were assessed in a mouse seizure model. The ED₅₀ against pentylenetetrazole-induced tonic seizures ranged from 0.48 mg/kg (L-676,893) to >160 mg/kg (L-685,869) cf. 0.26 mg/kg for diazepam. Although avermectins are without acute toxic effects, they have been historically shown to have relative low LD₅₀ values in mammals. The mechanisms involved in the anticonvulsant effect and the toxicity were investigated. A series of avermectin analogs displaced [³H]ivermectin binding to rat brain membranes and recombinant GABA_A receptors (α1β3γ2-subtype) with the same affinities, strongly suggesting that [³H]ivermectin labels the GABA_A receptor in rodent brain. Avermectins, which were anticonvulsant, were also potent inhibitors of [³H]ivermectin binding in rat brain. However, the rank order for anticonvulsant activity did not parallel the rank order for affinity at the [³H]ivermectin site and it was reasoned that avermectins may have differential affinity or efficacy at subtypes of the GABA_A receptor. All the active compounds tested potentiated the effects of GABA at recombinant GABA_A receptors in oocytes and at native cortical GABA_A receptors and the efficacy of avermectins at the GABA_A receptor correlated best with their anticonvulsant potency. Although avermectins weakly inhibited [³H]strychnine binding in rat spinal cord, and inhibited glycine responses on primary cultured cortical neurons, activity at glycine receptors did not correlate with either anticonvulsant activity or toxicity. Because both anticonvulsant activity and toxicity correlated best with activity at GABA_A receptors, it is unlikely that these effects can be separated, which may contraindicate the potential use of avermectins as anticonvulsants.