OBJECTIVES: In April 2005, the Food and Drug Administration (FDA) issued a black box warning (BBW) regarding the risks of using atypical antipsychotics (AAPs) for behavioral disorders in elderly patients with dementia. The objective of the present study was to investigate the impact of the BBW on the utilization of AAPs in elderly non-institutionalized dementia population. METHODS: Medical Expenditure Panel Surveys, from year 2004 through 2007, were used in the study. Utilization rates of AAPs pre-warning (2004-2005) and post-warning (2006-2007) were measures of primary interest in both overall and Medicare samples. Chi-square tests and multivariate logistic regression analyses were performed to examine pre-post differences in utilization rates (defined as elderly patients taking at least one medication associated with dementia) and gain insights into patterns of AAP use with respect to demographic, insurance and other factors. RESULTS: AAP use declined in the post-warning period for the main sample (X2= 0.81, p<0.3668) as well as Medicare sample (X2= 2.72, p<0.0992). However the decline was not statistically significant. Additionally, bivariate analyses of the main sample showed significantly higher proportion of patients receiving AAPs in the post-warning period for individuals with prescription drug coverage (x2= 21.63, p <0.0001). In contrast, a similar analysis for the Medicare sample revealed significantly higher proportion of patients receiving AAPs in the post-warning period for people with no Medicare drug coverage (x2= 20.83, p <0.0001).  Further, logistic regression analyses showed no significant decline in the use of AAPs in the post-warning period for the main sample (Odds Ratio, OR = 0.831, CI = 0.366 – 1.885) as well as the Medicare sample (OR= 0.798, CI= 0.364 – 1.753).  CONCLUSIONS: The regulatory warnings and labeling changes regarding off-label use of AAPs in dementia treatment seem to have made little impact on the actual use in non-institutionalized populations.