It has been argued that the occurrence of strike-slip and extensional earthquakes, as well as the presence of normal faults, in the high Andes of Peru, Bolivia, and northern Chile are a result of lateral changes in buoyancy stresses. Yet, no clear signal of extension is present in the published GPS velocities, suggesting that the entire overriding plate is presently under compression. This apparent ambiguity leads to the following questions: When do stress rates associated with the loading of normal faults operate? What are the dominant sources contributing to the horizontal stress field that can explain the distribution of seismicity? What are the magnitudes of the stresses involved? To what extent is the elastic loading at the trench contributing to the observed seismicity distribution? Is normal faulting in the high Andes time-dependent? We attempt to address these questions by quantifying the differences between the present-day and long-term horizontal velocity and strain rates fields, and we compare results with an obtained long-term dynamic model for the central Andes. We determine a present-day kinematic solution through a least-squares fit between model velocities and GPS vectors and by using the style and direction (not magnitude) of the seismic moment tensors as an a priori constraint on the style of the strain rate field. Due to the relatively large uncertainty in some of the GPS vectors, several different present-day models with distinct distributions of the elastic deformation are explored. A long-term kinematic solution is obtained through a least-squares fit to strain rates inferred from geologic data with imposed Nazca-South America velocity boundary conditions. We perform dynamic calculations to estimate the directions and magnitude of the long-term horizontal stress field. The sum of stress fields associated with gravitational potential energy differences (GPE) and the interaction of the Nazca and South American plates yield a total stress field. Estimates of GPE are inferred from topographic and geoid data. The relative importance of stresses related to GPE differences and the accommodation of relative plate motion is evaluated on the basis of a best fit between the total stress field and the style of faulting throughout the deformation zone. Using the long-term kinematic solution described above we determine a vertically averaged effective viscosity field. Forward calculations provide a dynamic long-term flow field for the region that is compared with the kinematic solutions.