There is no single “best” antipsychotic for dementia, because no antipsychotic is both universally effective and safe for this population. All antipsychotics carry an FDA black box warning noting that elderly patients with dementia-related psychosis face 1.6 to 1.7 times the risk of death compared to those taking a placebo. That said, one medication now has a specific FDA approval for dementia-related agitation, and several others are used off-label when behavioral symptoms become severe. The right choice depends on which symptoms are most disruptive, how the person tolerates side effects, and whether non-drug strategies have been tried first.
The Only FDA-Approved Option
In May 2023, the FDA approved brexpiprazole (Rexulti) as the first drug specifically indicated for agitation associated with dementia due to Alzheimer’s disease. The approval was based on two 12-week trials showing that patients taking 2 mg or 3 mg daily had statistically significant and clinically meaningful reductions in agitation compared to placebo, as measured by a standardized agitation rating scale.
This approval is narrow. It covers agitation in Alzheimer’s specifically, not psychosis, not other forms of dementia, and not the full spectrum of behavioral symptoms caregivers deal with. Still, it represents a shift: before this, every antipsychotic prescribed for dementia behaviors was technically off-label.
Common Off-Label Antipsychotics Compared
Three older atypical antipsychotics are the most frequently prescribed off-label for dementia-related behavioral symptoms: risperidone, olanzapine, and quetiapine. Each has a different profile of benefits and trade-offs.
Risperidone
Risperidone has the strongest evidence base for reducing aggression and psychosis in dementia. It’s typically started at a very low dose of 0.25 mg per day at bedtime, with a maximum around 2 to 3 mg daily in divided doses. The downside is a higher rate of movement-related side effects (stiffness, tremor, restlessness) compared to the other two options.
Olanzapine
A meta-analysis of over 2,400 participants found that olanzapine outperformed risperidone on several specific symptoms. It was more effective at reducing delusions and nighttime behavioral disturbances, and it carried statistically lower risks of agitation, sleep disturbance, and movement-related side effects. The overall response rate also favored olanzapine. It is typically started at 2.5 mg daily at bedtime, with a maximum of 10 mg per day. The main concern with olanzapine is weight gain and metabolic effects, including raised blood sugar and cholesterol, which need monitoring.
Quetiapine
Quetiapine is often chosen when sedation is actually a desired effect, such as for severe nighttime agitation or sundowning. It tends to cause fewer movement side effects than risperidone. Starting doses are low, around 12.5 mg twice daily, and can go up to 200 mg twice daily. The evidence for quetiapine’s effectiveness at reducing psychosis and aggression is generally weaker than for risperidone or olanzapine, but its milder side-effect profile makes it a common first try in frail or sensitive patients.
Why Non-Drug Approaches Come First
Guidelines from the American Psychiatric Association are clear: antipsychotics should only be used when behavioral symptoms are severe, dangerous, or causing significant distress, and only after non-drug strategies have been seriously attempted. The exceptions are major depression with suicidal thoughts, psychosis that puts someone at risk of harm, or aggression that threatens safety. In those cases, medication can start immediately.
Before reaching for a prescription, clinicians are expected to follow a structured process. The widely used DICE framework breaks this into four steps: thoroughly describing the behavior and its triggers, investigating reversible medical causes (urinary tract infections, constipation, pain, dehydration, poor sleep), creating a plan that prioritizes behavioral and environmental changes, and evaluating whether those changes worked. Many disruptive behaviors in dementia are driven by unmet needs, sensory deficits, or environmental confusion rather than by the disease itself.
Five practical strategies often reduce behavioral symptoms enough to avoid medication entirely: educating the caregiver about what’s driving the behavior, improving communication techniques, creating meaningful daily activities, establishing predictable routines with simplified tasks, and modifying the physical environment to reduce confusion and increase safety. These aren’t token suggestions. Research consistently shows they can meaningfully reduce agitation, aggression, and psychosis when applied consistently.
The Mortality Risk Is Real
The black box warning on every antipsychotic isn’t a formality. Across 17 placebo-controlled trials, elderly patients with dementia taking antipsychotics were 1.6 to 1.7 times more likely to die during the study period than those on placebo. The causes of death were varied but clustered around cardiovascular events (heart failure, sudden death) and infections, particularly pneumonia. This risk applies to all antipsychotics, both older and newer, and increases with higher doses and longer treatment.
This is why the decision to use an antipsychotic is always a calculated trade-off. When someone with dementia is punching caregivers, experiencing terrifying hallucinations, or refusing food because of paranoid delusions, the risk of the untreated symptoms may outweigh the risk of the medication. But the bar for starting treatment should be high, and the plan should always include a timeline for stopping.
How Long Treatment Should Last
Antipsychotics for dementia are meant to be short-term. The APA recommends that if there’s no meaningful improvement after four weeks at an adequate dose, the medication should be tapered and stopped. If the medication does work, an attempt to taper and withdraw it should happen within four months, unless previous tapering attempts led to a return of symptoms.
This recommendation exists because behavioral symptoms in dementia often fluctuate on their own. What looks like a medication success may actually be the natural waning of a behavioral episode. The only way to know is to carefully reduce the dose and see what happens. Many patients can be successfully tapered off without their symptoms returning.
Monitoring During Treatment
Anyone taking an antipsychotic for dementia needs regular monitoring. Within 30 days of starting the medication, baseline measurements should include weight, blood sugar (or hemoglobin A1c), and cholesterol. These same tests should be repeated around three months later. Antipsychotics can cause metabolic changes that raise the risk of diabetes and cardiovascular disease, and older adults with dementia are already vulnerable to both.
Caregivers should also watch for excessive sedation, falls, difficulty swallowing, and any new stiffness or unusual movements. These side effects can be subtle in someone who already has cognitive impairment, which makes regular check-ins with the prescribing clinician essential throughout treatment.