In a randomized controlled trial, the effectiveness of Parent–Child Interaction Therapy (PCIT) and correlates of maltreatment outcomes were examined. Mothers (N = 150) had a history or were at high risk of maltreating their children. After 12 weeks and compared to waitlist, PCIT mothers were observed to have improved parent–child interactions and reported better child behavior and decreased stress. At PCIT completion, improvements continued and mothers reported less child abuse potential and had improved maternal sensitivity. Also, PCIT completers were less likely to be notified to child welfare than noncompleters. Finally, those families not notified post-PCIT showed greater reductions in child abuse potential and improvements in observed sensitivity during treatment. Implications for theory and practice are discussed.