BackgroundTooth loss is suggested to be associated with an increased risk of dementia in many studies. But the relationship between tooth loss and dementia is not yet fully understood. This systematic review and meta-analysis aimed to determine the relative effect of tooth loss on dementia risk.MethodsAn electronic search of PubMed, Scopus, Embase, and Web of Knowledge was conducted in March 2018 to identify relevant observational studies with the English language restriction. Studies were included if they assessed the relationship between tooth loss and risk of dementia. Study quality was detected by the modified Downs and Black scale. Odds risks (ORs) were pooled using a random-effects model in the crude model.ResultsThe literature search initially yielded 1574 articles, and 21 observational studies published between 1994 and 2017 were finally included for the analyses. The crude results with random-effects model showed that patients with multiple tooth loss had higher incidence of dementia (OR 2.62, 95% CI 1.90–3.61, P < 0.001, I2 = 90.40%). The association remained noted when only adjusted results were pooled from 18 studies (OR 1.55, 95% CI 1.41–1.70, P = 0.13, I2 = 28.00%). Meta-regression analysis showed that study design explained about 16.52% of heterogeneity in the crude model. The overall quality rating scores of studies ranged from 11 to 16.ConclusionsFindings from this review evidenced that tooth loss is positively associated with an increased risk of dementia in adults. Future well-designed longitudinal researches examining the direct and indirect relationship between tooth loss and dementia risk are encouraged.Electronic supplementary materialThe online version of this article (10.1186/s12888-018-1927-0) contains supplementary material, which is available to authorized users.