Stress observations are compared to predictions of driving force models for the South American plate. Stress observations have the potential to give detailed information about the forces friving the plates, in that sets of forces with the same net torque can produce quite different stress patterns. For a two-dimensional (vertical section) model of a plate there are three conditions for dynamical equilibrium: the net forces and the net torque about any point must all be zero. These conditions are used to estimate the magnitudes of forces as a function of age, with ridge push and slab pull as active forces, and plate drag and slab resistance as passive forces. The results are consistent with earlier published models, and indicate that the primary blanace for an oceanic plate is between slab pull and the resisting drag forces. We generalize the model for the possible forces driving the South American plate and compare with observed stress measurements and kinematics. The South American plate has a small slab pull component, amounting to about one-fifth the ridge push force, with the other components reasonably isolated geographically. Thus, the plate balance is largely between ridge push and plate drag. This gives an opportunity of resolving the effects of trench suction or collision at the subduction zones. Most remarkable is the nearly uniform E-W orientation of the regional stress field, SHmax directions, which is extensional in the high Andes; a second feature is the N-S orientations in the Amazon basin. The two-dimensional, horizontal stress patterns are compared for successive models of combinations of driving forces. Models with ridge push, slab pull at the Scotia and Caribbean arcs, and trench suction near the west coast, balanced by plate drag, produce similar stress patterns to those observed. The magnitude of a possible trench suction force is estimated using a model with an eddy between the Nazca slab and the oberlying South American plate. ¿ American Geophysical Union 1992