Data from genetically modified mice suggest that benzodiazepine (BDZ)-site agonists with improved selectivity for α2-subtype GABAA receptors (α2GABAAR) are potentially useful for the treatment of neuropathic pain. Subtype-selective compounds available for preclinical tests in rodents support this concept but have not been approved for human use, hindering proof-of-concept studies in patients. We recently proposed that N-desmethyl clobazam (NDMC), the main metabolite of the licensed BDZ clobazam (CBZ), is responsible for most of the antihyperalgesia observed in mice after CBZ administration. In order to assess a potentially favorable pharmacological profile of NDMC, we analyzed differences in the GABAAR subtype specificity of CBZ, NDMC and diazepam (DZP) in recombinant receptors. DZP and CBZ potentiated sedating α1GABAARs and antihyperalgesic α2GABAARs with similar efficacies, whereas NDMC preferred α2GABAARs over α1GABAARs across a wide concentration range. In vivo, DZP and NDMC reduced neuropathic pain at doses between 3 and 30 mg/kg. At these doses, DZP had strong locomotor sedating effects while NDMC caused no or only weak sedation. Sedative effects of NDMC became apparent when the action of NDMC was restricted to α1GABAARs. However, when GABAAR point-mutated mice were studied that allow the analysis of antihyperalgesia and sedation in isolation, we found that, compared to DZP, NDMC had a significantly improved therapeutic window, consistent with its more favorable α2/α1 in vitro activity ratio. Given that NDMC should share the safety profile of its parent compound CBZ, it should be well-suited for proof-of-concept studies in human volunteers or patients.