Several lines of evidence support the general view of the San Andreas fault system (SAFS) as a major lithospheric weakness in a generally transpressive plate margin setting. However, the influence of the weakness of the SAFS on the observed stress and deformation fields is not straightforward because factors such as interactions between the brittle upper crust and ductile lower crust, lateral fault strength variation, and the amount of convergence may all be important. The goal of this study is to model steady state deformation, relative fault-parallel velocity, and crustal stress orientations in central and northern California using realistic rheologies and boundary conditions. Using a simplified three-dimensional (3-D) finite element analysis in large strain, we model the SAFS in 2-D cross sections with no stress or strain variations along strike. We investigate the influence of different parameters such as the frictional properties of the fault zone and the adjacent crust, the viscous properties of the lower crust as determined by its thermal structure, and the thermal structure of the lithosphere. Our model appears to provide a good conceptual framework for some first-order aspects of actively deforming plate margins such as the SAFS. The following findings emerge from a variety of numerical experiments: (1) The difference in the manner of transpressive strain partitioning in central and northern California can be explained by different fault strengths and the manner in which heat flow varies with distance from the fault. (2) Only the combination of a weak fault (with an effective friction coefficient of ≈0.1) and a strong lateral heat flow variation predicts approximately correct stress directions in the crust adjacent to the SAFS. ¿ 2001 American Geophysical Union