ABT-089 [2-methyl-3-(2-(S)-pyrrolidinylmethoxy)pyridine dihydrochloride], a novel ligand at neuronal nicotinic acetylcholine receptors with reduced adverse effects and improved oral bioavailability relative to (-)-nicotine, was tested in a variety of cognitive tests in rats and monkeys. Administered acutely, ABT-089 only marginally improved the spatial discrimination water maze performance of septal-lesioned rats. However, more robust improvement (45% error reduction on the last training day) was observed when ABT-089 was administered continuously via subcutaneous osmotic pumps (minimum effective dose: 1.3 micromol/kg/day). Continuous infusion of (-)-nicotine produced comparable improvement in the spatial discrimination water maze performance of septal-lesioned rats, but a 40-fold higher dose of (-)-nicotine was required (62 micromol/kg/day). Continuous infusion of ABT-089 to aged rats enhanced spatial learning in a standard Morris water maze, as indexed by spatial bias exhibited during a probe trial conducted after 4 days of training, but not when they were subsequently trained in a two-platform spatial discrimination water maze. The compound induced a small impairment in young rats on the standard water maze, but not on the two-platform task. A probe trial conducted after additional training in the standard water maze revealed no age or drug effects. ABT-089 did not affect performance of either the aged or young rats during inhibitory (passive) avoidance training. Also, continuous infusion of ABT-089 did not affect responses to acoustic startle or prepulse inhibition of acoustic startle in young, aged or septal-lesioned rats and did not affect locomotor activity in either sham-lesioned or septal-lesioned rats. In monkeys, acute administration of ABT-089 modestly improved the delayed matching-to-sample performance of mature, adult monkeys and more robustly improved performance in aged monkeys. Improved performance in the aged monkeys was restricted to the longest delay intervals and was not accompanied by changes in response latencies.