Atypical antipsychotics will continue to be prescribed for the behavioral symptoms of dementia in the absence of more effective, better tolerated, and safer alternatives. The evidence base, although incomplete, suggests that modest treatment effect sizes are offset by risk of considerable adverse effects. How might this information be best applied to clinical practice? Non-pharmacologic strategies should be implemented in routine clinical practice. Placebo-controlled clinical trials of individual antipsychotic agents have historically reported high placebo response rates; CATIE-AD reported that the sum total of the risk/benefit equation of atypical antipsychotic therapy was no greater than that achieved by placebo. CATIE-AD was designed to study the effectiveness of atypical antipsychotic treatment in community dwelling patients with AD. It is uncertain whether the results can be generalized to the populations of dementia patients residing in nursing homes with more severe cognitive and behavioral impairment. There is some suggestion that nursing home patients with dementia complicated by severe behavioral symptoms, particularly agitation and aggression without accompanying psychosis, might achieve greater benefit from atypical antipsychotic treatment than patients with milder behavioral symptoms. The finding that dementia patients without psychosis may respond more robustly to antipsychotic treatment seems counterintuitive, but may support the hypothesis that the neurobiology of the "psychosis of AD" differs from the psychosis of schizophrenia or bipolar disease. Adverse effects associated with antipsychotic therapy should be aggressively monitored throughout therapy. Treatment-emergent sedation was associated with all of the atypical antipsychotics in CATIE-AD and is probably an important mediator of mortality risk in patients with dementia. Sedation exacerbates pre-existing cognitive impairment and increases the risk of complications such as aspiration pneumonia, so concomitant use of benzodiazepines should be discouraged or limited to short periods with careful observation.' Once initiated, the effectiveness and tolerability of antipsychotic therapy should be evaluated routinely. In Alzheimer's disease, the severity and frequency of behavioral symptoms often decreases as illness progresses. In a stable patient, it is prudent to attempt to taper and discontinue the antipsychotic after 2-8 months of therapy. Better understanding of the potential adverse effects of antipsychotic therapy has increased interest in the effects of the dementia-specific medications on behavioral symptoms. Reductions in neuropsychiatric symptoms have been reported from trials of individual cholinesterase inhibitors, memantine monotherapy, and memantine combined with donepezil in AD patients. Studies of small numbers of patients in open trials of cholinesterase inhibitors (donepezil, rivastigmine, galantamine) and one double-blind placebo controlled trial (rivastigmine) have reported varying degrees of improvement of behavioral symptoms and psychosis of dementia with Lewy bodies (DLB). Delusions, hallucinations, apathy, and agitation/aggression are cited as the symptom categories most likely to show significant improvement. Since few of these studies were prospectively designed to study behavioral symptoms, results must be interpreted cautiously. Treatment of behavioral symptoms in AD and other dementias is challenging. The limitations of current approaches drive the search for effective, well tolerated therapies.