Mediational analyses were conducted with data from two small randomized control trials of the Responsive Teaching (RT) parent-mediated developmental intervention which used nearly identical intervention and control procedures. The purpose of these analyses was to determine whether or how the changes in maternal responsiveness and children's engagement that occurred during these interventions contributed to child development outcomes. To conduct these analyses, regression procedures described by Baron and Kenny were used to test for mediation. RT, which focuses on enhancing mother's responsiveness with their children, was found to be more effective than the control condition at promoting not only children's development but also both mothers' responsiveness and children's engagement (i.e., pivotal behavior). Intervention changes in mothers' responsiveness were highly associated with changes in children's pivotal behavior. However, intervention changes in children's language and personal social development were mediated by changes in children's pivotal behavior but not by changes in mothers' responsiveness. Results are discussed in terms of both mechanisms for developmental change and the nature of children's early intervention objectives.