In this study, an infant behavior control model is developed to prevent infant injuries by changing environmental factors. Infant climbing behaviors that can lead to fall injuries are the primary focus of the control model. Based on observed behavioral data, the probabilistic causal relationships between object attributes, infant characteristics, and climbing behaviors were modeled as a Bayesian Network Model. It was demonstrated that infant climbing behaviors could be predicted by the object attributes and controlled by changing object designs. It was also shown that infant injuries could be controlled by integrating both the developed behavior control model and the explanatory model of injuries that was constructed from infant injury data.