Evidence suggests that curcumin, the phytochemical agent in the spice turmeric, might be a potential therapy for Alzheimer's disease (AD). Its antioxidant, anti-inflammatory properties have been investigated extensively. Studies have also shown that curcumin can reduce amyloid pathology in AD. The underlying mechanism, however, is complex and is still being explored. In this study, we used the APPswe/PS1dE9 double transgenic mice, an AD model, to investigate the effects and mechanisms of curcumin in the prevention and treatment of AD. The water maze test indicated that curcumin can improve spatial learning and memory ability in mice. Immunohistochemical staining and Western blot analysis were used to test major proteins in β-amyloid aggregation, β-amyloid production, and β-amyloid clearance. Data showed that, 3 months after administration, curcumin treatment reduced Aβ40, Aβ42, and aggregation of Aβ-derived diffusible ligands in the mouse hippocampal CA1 area; reduced the expression of the γ-secretase component presenilin-2; and increased the expression of β-amyloid-degrading enzymes, including insulin-degrading enzymes and neprilysin. This evidence suggests that curcumin, as a potential AD therapeutic method, can reduce β-amyloid pathological aggregation, possibly through mechanisms that prevent its production by inhibiting presenilin-2 and/or by accelerating its clearance by increasing degrading enzymes such as insulin-degrading enzyme and neprilysin. © 2013 Wiley Periodicals, Inc.