Studies with our screening lead 5 and the literature compound 6 led to the identification of 6-benzyloxy-3-(4-methoxy)phenyl-1,2,4-triazolo[3,4-a]phthalazine 8 as a ligand with binding selectivity for the gamma-aminobutyric acid-A (GABA-A) alpha 3- and alpha 5-containing receptor subtypes over the GABA-A alpha 1 subtype (K(i): alpha 2 = 850 nM, alpha 3 = 170 nM, alpha 5 = 72 nM, alpha 1 = 1400 nM). Early optimization studies identified the close analogue 10 (K(i): alpha 2 = 16 nM, alpha 3 = 41 nM, alpha 5 = 38 nM, alpha 1 = 280 nM) as a suitable lead for further study. High-affinity ligands were identified by replacing the 6-benzyloxy group of compound 10 with 2-pyridylmethoxy (compound 29), but binding selectivity was not enhanced (K(i): alpha 2 = 1.7 nM, alpha 3 = 0.71 nM, alpha 5 = 0.33 nM, alpha 1 = 2.7 nM). Furthermore, on evaluation in xenopus oocytes,(22) 29 was discovered to be a weak to moderate inverse agonist at all four receptor subtypes (alpha 1, -7%; alpha 2, -5%; alpha 3, -16%; alpha 5, -5%). Replacement of the 3-phenyl group of 29 with alternatives led to reduced affinity, and smaller 3-substituents led to reduced efficacy. Methyl substitution of the benzo-fused ring of 29 at the 7-, 8-, and 10-positions resulted in increased efficacy although selectivity was abolished. Increased efficacy and retention of selectivity for alpha 3 over alpha 1 was achieved with the 7,8,9,10-tetrahydro-(7,10-ethano)-phthalazine 62. Compound 62 is currently one of the most binding selective GABA-A alpha 3-benzodiazepine-site partial agonists known, and although its selectivity is limited, its good pharmacokinetic profile in the rat (33% oral bioavailability after a 3 mg/kg dose, reaching a peak plasma concentration of 179 ng/mL; half-life of 1 h) made it a useful pharmacological tool to explore the effect of a GABA-A alpha 2/alpha 3 agonist in vivo.