Prior qualitative literature reviews about the use of neuroleptics in dementia suggest that they are "modestly effective" in treating agitation and that no single neuroleptic is better than another. To develop a more precise concept of the clinical efficacy of neuroleptics, a metaanalytic review of the existing literature was performed. From double-blind clinical trials that compared a neuroleptic with a placebo in agitated dementia patients, P values and effect-size estimates were obtained, and were assessed by metaanalytic techniques. Results indicated that neuroleptics were significantly more effective than placebo (one-tailed P = .004) and had a small effect size (r = .18). Clinically, neuroleptic treatment changed the improvement rate in agitated dementia patients from .41 to .59 (binomial effect-size display). This indicates that 18 of 100 dementia patients benefited from neuroleptic treatment (beyond that of placebo) and is consistent with the modest efficacy described in previous qualitative reviews. In six studies comparing thioridazine with another neuroleptic, and in five studies comparing haloperidol with another neuroleptic, metaanalysis results did not show that these two medications differed significantly from the comparison medications, which is not inconsistent with the opinion that no single neuroleptic is better than another.