We have constructed a simple dislocation model of the San Andreas fault near Cholame, California, that reproduces the observed deformation of nearby geodetic networks as well as the transition from quasi-continuous slip (fault creep) to a no-slip (locked) condition on the surface trace. The model is specified by the slip rates imposed on the fault surface. Slip on the surface trace is given by the observed fault creep rates, and slip at depths greater than 12.6 km is taken to 30 mm/yr along the entire length of the fault (i.e., beneath creeping and locked sections alike). Slip at intermediate depths is assigned by a rather arbitrary extrapolation of the surface creep values. This postulated distribution of slip on the fault is tested by using it in a dislocation model to predict the deformation in two geodetic networks, a trilateration network (20-km aperture), and a 20-km-long level line. The calculated deformation of the trilateration network agrees with the observed deformation within the observational error. Agreement between the predicted and observed elevation changes along the level line was satisfactory in the sense that the predicted values were below the level of detection, and the observed changes were not significant.