Pharmacotherapy of aggressive or agitated behaviors in the dementia patient has not been studied extensively, despite the prevalence of this problem. Neuroleptics have the most support for efficacy, with shorter acting benzodiazepines demonstrating benefit on occasion. However, studies done to date indicate that these drugs are effective only for a minority of patients and that side effects frequently make patients worse. Other medications, such as propranolol, carbamazepine, or lithium, may be helpful, but their efficacy in dementia patients has not been demonstrated in placebo-controlled studies. Until such studies are done, their use is most appropriate in special patient groups or in patients who have failed neuroleptic or benzodiazepine treatment. More studies are needed in elderly patients evaluating effectiveness of pharmacologic agents in specific types of dementia, particularly Alzheimer's disease. Most studies done to date have been of relatively short duration, usually two months or less. Because these medications often are given to dementia patients for prolonged periods, studies are needed to define the long-term clinical efficacy of these agents. In the clinical setting, these agents should be reduced periodically or discontinued to determine ongoing need. In addition, environmental, social, or behavioral methods of reducing agitated behaviors need to be explored as an adjunct to any medication trial.