The objective of this study was to assess the clinical evidence for or against mood stabilizers as a treatment for Alzheimer's disease (AD). We searched 5 databases from their inception to January 2010. Five randomized clinical trials of mood stabilizers to treat human patients suffering from AD were included. These trials assessed the effectiveness of mood stabilizers as an adjunct treatment to conventional anti-dementia drugs on behavioral and psychological symptoms, especially on agitation. Methodological quality was assessed using the Jadad score. The results suggested a significant effect in favor of placebo on the Mini-Mental Status Examination [n=270, weight mean difference (WMD), -0.89; 95% confidence intervals (CIs) -1.69 to -0.09, P=0.03] and on the Neuropsychiatric Inventory total (NPI total) (n=51, WMD, 3.71; 95% CIs 0.15 to 7.26, P=0.04). There were no significant differences in change scores on total Brief Psychiatric Rating Scale (BPRS total), NPI/BPRS agitation, Cohen-Mansfield Agitation Inventory total and Physical Self Maintenance Scale between mood stabilizers and placebo. Only one of these studies was free of methodological limitations (Jadad score=5). In conclusion, based on the existing evidence, mood stabilizers are ineffective or even harmful as a treatment for AD.