This paper proposes a “non-contact virtual clay modeling interface.” We developed a prototype of a three-dimensional modeling system that allows the user to deform the “virtual clay” by his or her hand movements. In our proposing method, the user’s hand movement is observed using multiple cameras to estimate its pose. The virtual clay is modeled by Catmull-Clark subdivision surface. Using the estimated hand pose, the virtual clay is deformed based on the direct free-form deformation technique. The developed prototype system indicates the feasibility of the proposed method for intuitive virtual clay modeling system, although it requires faster implementation for practical applications.