Theoretically, interventions initiated with at-risk infants prior to the point in time a definitive autism spectrum disorder (ASD) diagnosis can be made will improve outcomes. Pursuing this idea, we tested the efficacy of a parent-mediated early intervention called Adapted Responsive Teaching (ART) via a randomized controlled trial with 87 one-year-olds identified by community screening with the First Year Inventory as at-risk of later ASD diagnoses. We found minimal evidence for main effects of ART on child outcomes. However, ART group parents showed significantly greater increases in responsiveness to their infants than control group parents. Further, significant indirect (mediation) effects of assignment group on multiple child outcomes through changes in parent responsiveness supported our theory of change.