BACKGROUND: Acupuncture is often used as a treatment for dementia and is claimed to be effective in improving intelligence. AIMS: The objective of this review is to assess the clinical evidence for or against acupuncture as a treatment for Alzheimer's disease (AD). METHODS: We searched the literature using 17 databases from their inception to August 2008, without language restrictions. We included all randomised clinical trials (RCTs) of needle acupuncture to treat human patients suffering from AD. Methodological quality was assessed using the Jadad score. RESULTS: Three RCTs met all inclusion criteria. Two RCTs assessed the effectiveness of acupuncture on cognitive function compared with drug therapy. Their results suggested no significant effect in favour of acupuncture [n = 72, weight mean difference (WMDs), -0.55; 95% confidence intervals (CIs) -1.31 to 0.21, p = 0.15, heterogeneity: tau(2) = 0, chi(2) = 0.048, p = 0.49, I(2) = 0%]. Two RCTs tested acupuncture for activities of daily living (ADL). One RCT reported favourable effects of drug therapy compared with acupuncture for ADL, while the other failed to so. The meta-analysis of these data showed significant effects of drug therapy compared with acupuncture (n = 72, WMD, -1.29; 95% CIs: -1.77 to -0.80, p < 0.001, heterogeneity: tau(2) = 0, chi(2) = 0.17, p = 0.68, I(2) = 0%). CONCLUSION: Even though the number of studies is small, the existing evidence does not demonstrate the effectiveness of acupuncture for AD.