In regions of oblique plate convergence, deformation is generally partitioned into strike-slip and contractional components. We use a strain model of transpression, based upon a three-dimensional velocity gradient tensor, to address the question of why such partitioning occurs. An exact relationship between angle of plate convergence, instantaneous strain, and finite strain is calculated, providing a predictive tool to interpret the type and orientation of geological structures in zones of oblique convergence. The instantaneous and finite strain axes are not coincident, and thus there is no simple relationship between instantaneous strain (or stress) and structures that developed over a protracted deformation, such as crustal-scale faults. In order to simulate more realistic geological settings, we use a partitioning model of transpression which quantifies the effect of fault efficiency on the displacement field. This model is applied to two transpressional settings: Sumatra and central California. Sumatra shows a very low degree of partitioning of the displacement field and a relatively small offset on the Great Sumatran fault. In contrast, central California displays a large degree of displacement partitioning, and efficient slip on the San Andreas fault system. Since both systems show partitioning of deformation into strike-slip and thrust motion, it is unlikely that the San Andreas fault is rheologically 'weaker' than the Great Sumatran fault. We propose that the difference in fault efficiency results from kinematic partitioning of the displacement field controlled by relative plate motion, rather than mechanical decoupling generated by fault weakness.