The potential of amyloid-β (Aβ) immunization as a disease-modifying therapy for Alzheimer's disease is limited by the occurrence of encephalitic side effects in a subset of treated patients. The encephalitis was not predicted from immunization studies in transgenic, Aβ-depositing mice. More recently, studies in these same mice indicate that passive immunization with certain anti-Aβ antibodies can induce microhemorrhage. Cerebral amyloid angiopathy (CAA) may play a key role in determining the risk for these complications. Because aged nonhuman primates (NHPs) have a more human-like immune system than rodents, and because NHPs naturally develop senile plaques and CAA with age, NHPs appear to be important, adjunctive models for assessing the efficacy and safety of immunotherapeutics for Alzheimer's disease. Conversely, the ability to model the complications of Aβ immunotherapy will be important for elucidating the bases of these complications, and for developing protocols that minimize or eliminate the risks of these serious adverse effects.