The atypical antipsychotics (AAP), also known as second generation antipsychotics (SGAs) and serotonin–dopamine antagonists (SDAs), are a group of antipsychotic drugs (antipsychotic drugs in general are also known as tranquilizers and neuroleptics, although the latter is usually reserved for the typical antipsychotics) largely introduced after the 1970s and used to treat psychiatric conditions. Some atypical antipsychotics have received regulatory approval (e.g. by the FDA of the US, the TGA of Australia, the MHRA of the UK) for schizophrenia, bipolar disorder, irritability in autism, and as an adjunct in major depressive disorder.

Both generations of medication tend to block receptors in the brain's dopamine pathways. Atypicals are less likely than haloperidol—the most widely used typical antipsychotic—to cause extrapyramidal motor control disabilities in patients such as unsteady Parkinson's disease–type movements, body rigidity, and involuntary tremors. However, only a few of the atypicals have been demonstrated to be superior to lesser-used, low-potency first-generation antipsychotics in this regard.

As experience with these agents has grown, several studies have questioned the utility of broadly characterizing antipsychotic drugs as "atypical/second generation" as opposed to "first generation", noting that each agent has its own efficacy and side-effect profile. It has been argued that a more nuanced view in which the needs of individual patients are matched to the properties of individual drugs is more appropriate. Although atypical antipsychotics are thought to be safer than typical antipsychotics, they still have severe side effects, including tardive dyskinesia (a serious movement disorder), neuroleptic malignant syndrome, and increased risk of stroke, sudden cardiac death, blood clots, and diabetes. Significant weight gain may occur. Critics have argued that "the time has come to abandon the terms first-generation and second-generation antipsychotics, as they do not merit this distinction."

Atypical antipsychotics are typically used to treat schizophrenia or bipolar disorder. They are also frequently used to treat agitation associated with dementia, anxiety disorder, autism spectrum disorder, persecutory delusion and obsessive-compulsive disorder (an off-label use). In dementia, they should only be considered after other treatments have failed and if the patient is a risk to themselves or others.

The first-line psychiatric treatment for schizophrenia is antipsychotic medication, which can reduce the positive symptoms of schizophrenia in about 8–15 days. Antipsychotics only appear to improve secondary negative symptoms of schizophrenia in the short term and may worsen negative symptoms overall. Overall there is no good evidence that atypical antipsychotics have any therapeutic benefit for treating the negative symptoms of schizophrenia.

There is very little evidence on which to base a risk and benefit assessment of using antipsychotics for long-term treatment.

The choice of which antipsychotic to use for a specific patient is based on benefits, risks, and costs. It is debatable whether, as a class, typical or atypical antipsychotics are better. Both have equal drop-out and symptom relapse rates when typicals are used at low to moderate dosages. There is a good response in 40–50% of patients, a partial response in 30–40%, and treatment resistance (failure of symptoms to respond satisfactorily after six weeks to two of three different antipsychotics) in the remaining 20%. Clozapine is considered a first choice treatment for treatment resistant schizophrenia, especially in the short term; in the longer-terms the risks of adverse effects complicate the choice. In turn, risperidone, olanzapine, and aripiprazole have been recommended for the treatment of first-episode psychosis.

The utility of broadly grouping the antipsychotics into first generation and atypical categories has been challenged. It has been argued that a more nuanced view, matching the properties of individual drugs to the needs of specific patients is preferable. While the atypical (second-generation) antipsychotics were marketed as offering greater efficacy in reducing psychotic symptoms while reducing side effects (and extrapyramidal symptoms in particular) than typical medications, the results showing these effects often lacked robustness, and the assumption was increasingly challenged even as atypical prescriptions were soaring. In 2005 the US government body NIMH published the results of a major independent (not funded by the pharmaceutical companies) multi-site, double-blind study (the CATIE project). This study compared several atypical antipsychotics to an older, mid-potency typical antipsychotic, perphenazine, among 1,493 persons with schizophrenia. The study found that only olanzapine outperformed perphenazine in discontinuation rate (the rate at which people stopped taking it due to its effects). The authors noted an apparent superior efficacy of olanzapine to the other drugs in terms of reduction in psychopathology and rate of hospitalizations, but olanzapine was associated with relatively severe metabolic effects such as a major weight gain problem (averaging 9.4 lbs over 18 months) and increases in glucose, cholesterol, and triglycerides. No other atypical studied (risperidone, quetiapine, and ziprasidone) did better than the typical perphenazine on the measures used, nor did they produce fewer adverse effects than the typical antipsychotic perphenazine (a result supported by a meta-analysis by Leucht et al. published in The Lancet), although more patients discontinued perphenazine owing to extrapyramidal effects compared to the atypical agents (8% vs. 2% to 4%, P=0.002). A phase 2 part of this CATIE study roughly replicated these findings. Compliance has not been shown to be different between the two types. Overall evaluations of the CATIE and other studies have led many researchers to question the first-line prescribing of atypicals over typicals, or even to question the distinction between the two classes.