A new working hypothesis that there is a hitherto unrecognized binding site on the aldose reductase (AR) enzyme with strong affinity for benzothiazoles was pursued for the design of novel, potent aldose reductase inhibitors (ARIs). The first application of this hypothesis led to a novel series of 3,4-dihydro-4-oxo-3-(benzothiazolylmethyl)-1-phthalazineacetic+ + + acids. The parent of this series (207) was a potent inhibitor of AR from human placenta (IC50 = 1.9 x 10(-8) M) and was orally active in preventing sorbitol accumulation in rat sciatic nerve, in an acute test of diabetic complications (ED50 = 18.5 mg/kg). Optimization of this lead through medicinal chemical rationale, including analogy from other drug series, led to more potent congeners of 207 and culminated in the design of 3,4-dihydro-4-oxo-3-[[5-(trifluoromethyl)-2-benzothiazolyl] methyl]-1-phthalazineacetic acid (216, CP-73,850, zopolrestat). Zopolrestat was found to be more potent than 207, both in vitro and in vivo. Its IC50 against AR and ED50 in the acute test were 3.1 x 10(-9)M and 3.6 mg/kg, respectively. Its ED50s in reversing already elevated sorbitol accumulation in rat sciatic nerve, retina, and lens in a chronic test were 1.9, 17.6, and 18.4 mg/kg, respectively. It was well absorbed in diabetic patients, resulting in high blood level, showed a highly favorable plasma half-life (27.5 h), and is undergoing further clinical evaluation. An assortment of synthetic methods used for the construction of benzothiazoles, including an efficient synthesis of zopolrestat, is described. Structure-activity relationships in the new series are discussed.