This study investigated the association between colonization of mother-child dyads with Streptococcus mutans (SM) and early childhood caries (ECC), with consideration of the mediator and confounders, using a theoretical model. Four hundred mothers and their children from São Luís, Brazil, were included in the study. A diagram based on directed acyclic graphs was elaborated to analyze the association between SM colonization of the mother, mediated by child SM, and ECC. Other maternal (socioeconomic, waist circumference, sugar consumption, DMFT index, and visible plaque) and child factors (sugar consumption, visible plaque, and child age) composed the theoretical model. A total effect model (maternal SM on ECC) and a direct effect model, adjusted by the mediation effect of child SM (maternal SM via child SM) on ECC, were analyzed. The outcome of interest in this study was ECC (number of teeth with caries experience, dmft), which was analyzed as a discrete variable by Poisson regression. The Paramed test was used to analyze mediation. The following variables were associated with ECC in the total effect model: maternal SM, maternal waist circumference ≥80 cm, DMFT, maternal visible plaque, child age ≥4 years, and increased sugar consumption of the child (>3 times/day). In the direct effect model, high maternal SM levels remained associated with ECC, while moderate and high colonization of the child with SM was also associated with ECC. Child SM colonization just partially mediated the effect of maternal SM on ECC (33%). Thus, factors other than the mechanism of bacterial colonization should be considered in mother-child dyads, including eating habits, oral hygiene practices, and a family history of caries.