Mitochondrial DNA (mtDNA) content alterations are potential mediators of neuronal disorders. However, what effects of mtDNA content are on fluoride-induced neurotoxicity remain elusive. Here, a cross-sectional study was conducted to investigate roles of circulating mtDNA content on associations between fluoride and intelligence quotient (IQ) in 605 children aged 7–13 years from Tianjin, China. We assessed IQ and circulating mtDNA content in peripheral blood through the second edition of Combined Raven’s Test-The Rural in China (CRT-RC2) and qRT-PCR methods. National standardized ion selective electrode method was used for fluoride concentrations in drinking water samples from each village and urine samples. Our results showed that the median (interquartile range) of water and urinary fluoride concentrations were 1.00 (0.70–2.20) mg/L and 0.41 (0.15–2.29) mg/L respectively. Water/urinary fluoride concentrations were negatively associated with not only IQ scores but mtDNA content, which in turn was positively linked with odds of children having excellent intelligence (P for trend < 0.05). In gender-stratified analysis, fluoride levels were correlated with decreased odds [odds ratio (OR) = 0.94, 95% confidence interval (CI): 0.90-0.99 for water fluoride; OR = 0.94, 95% CI: 0.89-0.99 for urinary fluoride], while mtDNA content was related to increased odds of having excellent intelligence (OR = 1.18; 95% CI: 1.07-1.30) in girls. Mediation analyses indicated that mtDNA content mediated 38.2% (95% CI: 9.6-78.4%) and 29.2% (95% CI: 0.77-67.0%) of associations of fluoride in water and urine with reduced odds of having excellent intelligence in girls. Collectively, our findings suggest that circulating mtDNA content was negatively associated with low-moderate fluoride concentrations, whereas positively linked to odds of possessing excellent intelligence in school-age children. Especially, circulating mtDNA content may mediate the association between fluoride and reduced odds of having excellent intelligence in girls. Further, our findings need to be verified in a larger and general population.