Various alpha(2,3) subtype selective partial GABA-A agonists are in development to treat anxiety disorders. These compounds are expected to be anxiolytic with fewer undesirable side effects, compared to nonselective GABA-A agonists like benzodiazepines. Several alpha(2,3) subtype selective and nonselective GABA-A agonists have been examined in healthy volunteers, using a battery addressing different brain domains. Data from five placebo-controlled double-blind studies were pooled. Lorazepam 2 mg was the comparator in three studies. Three alpha(2,3)-selective GABAA agonists (i.e., TPA023, TPACMP2, SL65.1498), one alpha(1)-selective GABAA agonists (zolpidem), and another full agonist (alprazolam) were examined. Pharmacological selectivity was assessed by determination of regression lines for the change frombaseline of saccadic-peak-velocity-(Delta SPV-) relative effect, relative to changes in different pharmacodynamic endpoints (Delta PD). SPV was chosen for its sensitivity to the anxiolysis of benzodiazepines. Slopes of the Delta SPV-Delta PD relations were consistently lower with the alpha(2,3) selective GABA-A agonists than with lorazepam, indicating that their PD effects are less than their SPV-effects. The Delta SPV-Delta PD relations of lorazepam were comparable to alprazolam. Zolpidem showed relatively higher impairments in Delta PD relative to.SPV, but did not significantly differ from lorazepam. These PD results support the pharmacological selectivity of the alpha(2,3)-selective GABA-A agonists, implying an improved therapeutic window.