Self-injury is a devastating, maladaptive behavior disorder that is common in developmental disabilities and is comorbid with numerous psychiatric disorders. Examples of self-injurious behavior (SIB) include head-banging, self-biting, and self-punching. The neurochemical basis of SIB is unknown; however, many different classes of drugs are prescribed (e.g., neuroleptics, atypical neuroleptics, anti-epileptics, opioid antagonists) to reduce these behaviors. These drugs have all shown clinically significant but limited efficacy in patient populations, and no class of drug is effective for all patients. The development and characterization of a valid animal model could provide important information regarding the neurochemical basis of SIB and could be used to screen potential new pharmacotherapies. In one model of SIB, high doses of pemoline (2-amino-5-phenyl-1,3-oxazol-4-one) are administered to rats. Using this model, we evaluated the effectiveness of three drugs (risperidone, valproate, and topiramate) that reduce SIB in humans. We also screened the potential effectiveness of tramadol, a drug that decreases stereotyped and compulsive behaviors but has not been assessed in human self-injurers. We found that risperidone, valproate, and topiramate each significantly attenuate pemoline-induced SIB, whereas tramadol does not. These findings suggest that the pemoline model of SIB has predictive validity across a range of drug classes and implicate important potential neurochemical mechanisms that may contribute to the behavior disorder. The findings also indicate that tramadol may not be an effective pharmacotherapy for SIB.