The aim of this study was to assess the prevalence of long-term antipsychotic (AP) use in community-dwelling patients with dementia considering hospitalization periods as AP exposure or not. A retrospective study was carried out from 2009 to 2012 on a PACA-Alzheimer cohort (which included 31 963 patients in 2009 and 36 442 in 2012 from 5 million inhabitants). Three groups of patients were identified according to the longest exposure to APs without interruption: nonusers, short-term users (≤3 successive months without discontinuation), and long-term users. Sensitivity analyses on hospitalization periods were carried out. The percentage of patients with at least one AP dispensing was stable over the study period (25.6% in 2009 vs. 26.5% in 2012). In 2012, 27.6% were AP long-term users. This increased to 46.7% when hospitalization periods were counted as AP exposure. In comparison with nonusers, AP users took more benzodiazepines and antidepressants. Short-term users were men [odds ratio (OR)=1.2, 95% confidence interval (CI) (1.1-1.3)] older than 85 years old [OR=1.2, 95% CI (1.1-1.2)]. Long-term users were more exposed to benzodiazepines [OR=1.2, 95% CI (1.1-1.4)]. This study showed that long-term use of AP remained frequent in community-dwelling patients with dementia. It also showed that the prevalence of long-term users almost doubled when hospitalization periods were counted as AP exposure. This underlines the need to consider hospitalization periods when assessing medication exposure in populations with frequent periods of hospitalization.