Although randomized interventions trials have been shown to reduce the incidence of disorganized attachment, no studies to date have identified the mechanisms of change responsible for such reductions. Maternal sensitivity has been assessed in various studies and shown to change with intervention, but in the only study to formally assess mediation, changes in maternal sensitivity did not mediate changes in infant security of attachment (Cicchetti, Rogosch, & Toth, 2006). Primary aims of the current randomized controlled intervention trial in a high-risk population were to fill gaps in the literature by assessing whether the intervention (a) reduced disorganization, (b) reduced disrupted maternal communication, and (c) whether reductions in disrupted maternal communication mediated changes in infant disorganization. The results indicated that, compared to controls (n = 52), both infant disorganization and disrupted maternal communication were significantly reduced in the intervention group (n = 65) that received regular home-visiting during pregnancy and the first year of life. Furthermore, reductions in disrupted maternal communication partially accounted for the observed reductions in infant disorganization compared to randomized controls. The results are discussed in relation to the societal cost effectiveness of early attachment-informed interventions for mothers and infants, as well as the importance of formally assessing underlying mechanisms of change in order to improve and appropriately target preventive interventions.