Pharmacotherapy in patients with dementia aims to improve distressing behavioral and psychological signs of dementia after nonpharmacologic interventions fail, without causing unacceptable side effects or exacerbating underlying cognitive impairment. We review data describing risperidone (3 published placebo-controlled trials), olanzapine (1 abstract regarding a placebo-controlled trial and a published placebo-controlled trial), quetiapine (1 published open-label trial and an abstract regarding a placebo-controlled trial), and aripiprazole (1 abstract regarding a placebo-controlled trial). For example, a 12-week study of risperidone in patients with Alzheimer's disease showed a dose-related improvement in psychosis and agitation. The frequency of extrapyramidal symptoms (EPS) was also significantly greater in patients receiving the highest doses. A 6-week study of olanzapine showed greater improvement than placebo in agitation/aggression and psychosis with doses of 5 and 10 mg/day, but not 15 mg/day, with side effects including gait disturbance and sedation at all doses. A 52-week, open-label trial of quetiapine (median dose = 138 mg/day) in elderly patients with psychosis suggested good tolerability with apparent behavioral benefit; EPS improved or remained unchanged in most patients. Limited data describing aripiprazole have shown inconclusive evidence regarding relief of psychosis in elderly patients with Alzheimer's disease-related dementia, with apparently good tolerability over the short term. It appears that, in the aggregate, atypical antipsychotics are efficacious for treatment of agitation in dementia, with less clear impact on psychosis, but their tolerability profiles clearly differ. The National Institute of Mental Health-funded Clinical Antipsychotic Trials of Intervention Effectiveness in Alzheimer's Disease project will provide the first head-to-head comparisons of atypicals in dementia and will examine possible drug-drug differences between efficacy and effectiveness.