Atomoxetine is a selective norepinephrine (NE) reuptake blocker that has recently been marketed for the treatment of attention deficit hyperactivity disorder. The purpose of the present study was to evaluate the self-administration of atomoxetine in an animal model predictive of abuse liability in humans. Rhesus monkeys (N = 5) were prepared with chronic intravenous catheters and allowed to self-administer cocaine or saline during alternating baseline sessions. When behavior was stable, atomoxetine (0.03-3.0 mg/kg per injection), desipramine (0.1-3.0 mg/kg per injection), methylphenidate (0.001-0.1 mg/kg per injection), or their vehicles were substituted for baseline conditions. Methylphenidate consistently maintained responding above the levels maintained by its vehicle. Atomoxetine and desipramine failed to reliably maintain self-administration above vehicle levels in four of five individual monkeys. These results predict that atomoxetine, in contrast to methylphenidate but like desipramine, will lack reinforcing effects and abuse potential in humans.