Introduction: The neuroprotective effect of valproic acid has been observed in the animal models of neurodegeneration, which suggests it as a potential candidate for clinical trials. In this paper, we aimed to systematically analyze the efficacy and safety of valproic acid in the treatment of dementia. Methods: We searched the electronic databases PubMed, EMBASE, CINAHL, Cochrane Library and China National Knowledge Infrastructure until March 2020 for the eligible randomized controlled trials, as well as the unpublished and ongoing trials. We pooled the results using a random-effects model. Results: We included seven studies with 770 randomized patients with dementia, which compared valproic acid with placebo. Indeed, there were no significant differences found in the scores of Mini–mental State Examination, Cohen-Mansfield Agitation Inventory and number of patients with adverse events. Valproic acid is generally well-tolerated in patients with dementia, even in long-term therapy for 24 months. Conclusion: Insufficient evidences are found to support valproic acid in the treatment of dementia for cognitive, psychiatric symptoms or disease-modifying. The anticipations for a success in the trial of valproic acid for dementia in the future look not optimistic based on the available evidence