Agitation occurs commonly in patients with dementia. Before symptomatic pharmacotherapy is undertaken, it is imperative to perform a sequence of evaluations and interventions to establish whether simpler and safer, nonpharmacologic approaches will be beneficial. When psychotropic medications are used they should be used judiciously, in the lowest effective doses and for the shortest period of time necessary. Ineffective medications should be stopped, and even effective medications should be empirically tapered in most patients to learn whether treatment is still necessary. Antipsychotics probably show the greatest benefit for agitation associated with psychotic features; they have less demonstrated efficacy for agitation not associated with psychotic features. The side effects of typical agents are legion; data are pending regarding atypical agents. The available evidence regarding nonneuroleptic medications ranges from case reports to well-designed, double-blind, placebo-controlled, randomized, parallel group studies. Literature exists describing the use of anticonvulsants, anxiolytics, serotonergic antidepressants, and other agents to manage agitation. Carbamazepine and divalproex sodium (valproate) have demonstrated efficacy in uncontrolled studies, whereas the use of carbamazepine has produced negative results in one small controlled study and positive results in two larger controlled studies. Buspirone has shown benefit in some open trials. Encouraging early findings have been reported for trazodone, including from one controlled trial. Varying results have been obtained using selective serotonin reuptake inhibitors, but with consistently encouraging anecdotes. In the aggregate, the evidence suggests but does not prove that alternatives to traditional antipsychotics exist. Again, none of these agents has yet been approved for this purpose by the FDA. As more studies become available we will have a better idea about which classes of agents are most efficacious. It is likely that there may be a role for "rational" polypharmacy in the management of this distressing complication of dementia. However, no studies that we know of address combination therapy, so the clinician must contemplate this option on a case-by-case basis. Clinical trials data are pending from studies with divalproex sodium, carbamazepine, haloperidol versus trazodone versus placebo, risperidone, olanzapine, quetiapine, donepezil, xanomeline, tacrine, buspirone, and sertraline, at the very least. These data will undoubtedly have a major impact on how we care for our patients and lead to revisions of current practice guidelines.