The objectives of this retrospective, naturalistic study were to provide preliminary data on the effects of 6 months treatment with risperidone, olanzapine and quetiapine on behavioral disturbances, within a sample of outpatients with mild to moderate Alzheimer's disease, and on predictors of response. Between July 2005 and December 2005, data were collected from 58 consecutive outpatients with a DSM-IV-TR diagnosis of Alzheimer's disease with behavioral disturbances, who received a 6-month treatment with risperidone, olanzapine or quetiapine. Primary outcome measures were Neuropsychiatric Inventory (NPI) total score and its items forming the basic core of behavioral disturbances in Alzheimer's disease: delusions, hallucinations and agitation/aggressiveness. Secondary outcome measures were Mini-Mental State Examination (MMSE), Activities of Daily Living, Instrumental Activities of Daily Living and Clinical Insight Rating scale. Correlations between baseline MMSE score and improvements in behavioral disturbances were investigated. At 6 months mean NPI total score had fallen 43.5% in the risperidone group, 45.6% in the olanzapine group and 33.3% in the quetiapine group, with no significant between-group differences. Global cognitive function showed no significant change from baseline to end-point. Incidence of adverse events was low. A significant correlation was found between MMSE score and NPI total score and NPI item agitation decreases. Risperidone, olanzapine and quetiapine produced significant improvements in behavioral disturbances and were well tolerated. No significant differences emerged among treatments. The preliminary results also suggest that baseline cognitive function might influence treatment response.