Among a host of negative outcomes, exposure to early adversity is associated with dysregulation, including differences in diurnal cortisol (morning levels and slope), poorer emotion regulation, and lower inhibitory control. Past work examining cumulative effects of early adversity tend to employ a sum score approach whereby individual risk variables are dichotomized and summed. A composite latent variable approach to assessing early adversity improves on key limitations of the traditional sum score approach by allowing for differential weighting of individual factor variables, continuous and dichotomous indicator variables, and the simultaneous estimation of cumulative effects of early adversity and individual effects of each risk variable. The current study examined these two methodological approaches in a serial mediation model of developmental mechanisms theorized to explain links between early adversity at the family level and three markers of child self-regulation. Maternal sensitivity and supportive emotion-related socialization behaviors have been theorized to serve as developmental process variables that explain associations between early adversity and child self-regulation. Further, maternal sensitivity during parent-child interactions is expected to inform child internal working models, and in typically developing samples child representations of appropriate self-regulation have been linked with actual child self-regulation. However, little longitudinal work has examined theory and evidence-based developmental mechanisms that might explain associations between adversity and dysregulation using multiple indices of child self-regulation while also considering mediational roles of maternal sensitivity in multiple contexts and child internal working models. The current study examined longitudinal serial mediation using a model building approach employing 8 structural equation models including latent growth curve components whereby early adversity was hypothesized to predict three indices of child self-regulation (diurnal cortisol, emotion regulation, and inhibitory control) and their change across time, through influences of maternal sensitivity during free play and reminiscing on child internal working models of self-regulation. Data was drawn from a randomized controlled trial and included 229 mothers and their 3- to 6-year-old children across four measurement time points (baseline [T1], 2 months [T2], 6 months [ T3], and 1 year later [T4]). Results revealed unique effects of cumulative early adversity predicting increases in diurnal cortisol slope from T1 to T4 and lower child emotion regulation at T4, as well as lower maternal sensitivity during free play and lower maternal sensitive guidance during reminiscing at T2. Higher maternal sensitivity during free play at T2 predicted lower child representations of behavior dysregulation during child narratives at T3, and higher maternal sensitive guidance during reminiscing at T2 predicted higher representations of moral affiliative themes during child narratives at T3. Further, there was evidence of significant serial indirect effects of cumulative early adversity on lower child inhibitory control at T4 through lower maternal sensitive guidance at T2 and lower moral affiliative behavioral representations at T3. Results of the final serial mediation model were generally consistent when comparing the effects of cumulative early adversity using the sum score and the latent composite approaches. The latent compose approach revealed that low income status and maternal perpetration of neglect were the strongest contributors to the weighted early adversity composite. Results inform understanding of mechanisms that contribute to child self-regulation development during the preschool years among high risk children and families.