Despite the rising cultural phenomenon of grandparents parenting grandchildren on a full-time basis due to problems within the birth parent generation, intervention studies with these families have been scarce, methodologically flawed, and without conceptual underpinnings. We conducted a randomized clinical trial (RCT) with 343 custodial grandmothers recruited from across 4 states to compare the effectiveness of behavioral parent training (BPT), cognitive–behavioral therapy (CBT), and information-only control (IOC) conditions at lowering grandmothers’ psychological distress, improving their parenting practices, and reducing the internalizing and externalizing difficulties of target grandchildren between ages 4 and 12. These outcomes were derived conceptually from the family stress model and modeled as latent constructs with multiple indicators. Each RCT condition was fully manualized and delivered across 10 sessions within groups led jointly by trained professionals and peer facilitators in community settings. Multidomain second-order latent difference score models were performed on a full intent-to-treat basis to compare the 3 RCT conditions on changes in the above outcomes from baseline to postintervention and from baseline to 6 months postintervention. In general, while CBT and BPT interventions were both superior to IOC at both times of measurement on most outcomes, they differed little from each other. Effect sizes were generally in the moderate to large range and similar to those found in prior studies of BPT and CBT with traditional birth parents. We conclude from this research that evidence-based interventions focusing on appropriate skill development and behavioral change can yield positive outcomes within custodial grandfamilies.