Analog sandbox models provide cheap, concise data and allow the evolution of geological structures to be observed under controlled laboratory conditions. Seismic physical modeling is used to study the effects of seismic wave propagation in isotropic and anisotropic media, and to improve methods of data acquisition, processing and, interpretation. By combining these two independent modeling techniques, the potential exists to expand the benefits of each method. For seismic physical modeling, the main advantages are that the seismic data collected from these models contain natural variations that cannot be built into conventional solid models, which are machined with predetermined structures. In addition, the cost and construction time to build these models is significantly reduced. For sandbox modeling, the ability to record three-dimensional (3-D) seismic images before the model is manually sectioned for conventional two-dimensional (2-D) structural interpretation allows far more detailed study of subtle 3-D structures than previously possible. In the past other workers have attempted to use unconsolidated sands for seismic physical models, but these attempts have been unsuccessful because of the lack of control or understanding of the natural variations that occur throughout the models. The aim of our research has been to overcome such problems and to develop techniques to alter the acoustic impedance of sand layers, thereby allowing the subsequent ultrasonic seismic recording of 3-D fault systems in sand analog geological models.