Previous research has identified factors related to outcome in child anxiety treatment and parent training programs for child behavior problems. However, it is unclear what factors predict outcomes in interventions delivered online to parents of young children at risk of anxiety. This study investigated predictors of child anxiety outcomes among 433 families with young children (3-6 years) who participated in a randomized controlled trial of Cool Little Kids Online, an eight-module early intervention program for child anxiety based on cognitive-behavioral therapy (CBT). Potential predictors included baseline demographic factors, child and parent mental health factors, and indicators of program use, including number of online modules completed and frequency of homework practice. Results showed that only access to a printer moderated intervention effectiveness. Printer access predicted lower child anxiety in the Cool Little Kids Online group, but had no effect on outcomes in the wait-list group. In both groups, higher levels of child anxiety symptoms, child-inhibited temperament, and poorer parent mental health at baseline predicted higher levels of child anxiety symptoms at 6-month follow-up. The amount of online program use was not related to improvements in child anxiety symptoms. However, parents who reported practicing the program skills more frequently showed greater reductions in child anxiety, and access to a printer was related to frequency of program skills practice. These findings provide empirical support for the important role of skills practice in online CBT interventions, and suggest that practicing program skills may be more important than completing the online modules.