Background: antipsychotic drugs (APs) have been associated with falls and fractures in elderly individuals but limited knowledge on specific drugs exist. Objective: to investigate the association between individual APs and fractures in elderly persons. Design and setting: nationwide register-based cohort study. Subjects: all Danish individuals aged ≥65 who had not been in treatment with any AP in the year before inclusion. Methods: incidence rate ratios (IRRs) of fractures of hip, pelvis or upper extremities during treatment with commonly used APs were assessed in multivariable Poisson models. Exposure was divided into time periods from initiation of treatment: 0-30 days, 31-365 days or > 365 days. Results: one year prior to inclusion, 1,540,915 individuals =65 years had not received APs and of these 93,298 initiated treatment with APs. Mean follow-up was 9.6 years. During follow-up, 246,057 (16%) experienced a fracture. Associations were for all APs highest in the initial treatment period (0-30 days) with IRRs for risperidone 1.97 (95% CI: 1.70-2.28), olanzapine 2.31 (95% CI: 1.96-2.73), quetiapine 2.09 (95% CI: 1.73-2.52), zuclopenthixol 2.19 (95% CI: 1.82-2.63), chlorprothixen 1.62 (95% CI: 1.18-2.24), flupenthixol 1.43 (95% CI: 1.06-1.93), levomepromazine 1.19 (95% CI 0.86-1.66), haloperidol 2.98 (95% CI 2.57-3.45), compared with the background population. Conclusions: use of APs is associated with fractures in elderly persons especially in the initial treatment period. If AP use in an elderly person is deemed necessary, individual falls prophylaxis should be considered.