This paper presents a review of strain-stress relationships in the active Taiwan collision at different scales based on numerical modeling. As the digital computer has become an important research tool due to its capability of performing numerical simulations of real tectonic processes, a mass of accurate and comprehensive database is required to assure the meaning-fulness and accuracy of simulation results. The aim of this study is to refine the general interpretation of the relationships between active deformation and geological structures. Through an evaluation of the presence and rele of mechanical decoupling along major faults, distinctive attention will be devoted to the role of geological discontinuities. Such decoupling plays an important role in the distribution of regional and local stress and velocity patterns. In this paper, the relationship between the complexity of modeling and the data within the range of data uncertainties is also discussed in order to determine the level of complexity at which a model can be built, validated and considered significant. The construction of numerical simulation depends on the scale of the model and on the data available as constraints within that scale. The study of the Taiwna collision case provides an illustration of the relationships between kinematics, structure and/or strain fields in a curved belt. It is claimed that although 2-D numerical modeling provides valuable results, validation through 3-D modeling experiments is indispensable in accounting for the oblique dips of major boundaries.