Two sets of finite-element model suites were developed to address discrepancies between geodetic and geologic fault slip rates in southern California, as well as the extent of “off-fault” deformation that does not result from interseismic elastic strain around known, locked faults. The GPS-constrained models represent current deformation, while the strain energy rate (TSE) minimizing models represent a long-term average of deformation that has occurred over thousands to millions of years. Both classes of models suggest low slip rates (<29 mm/yr) for the San Andreas Fault (SAF) in the Mojave and Big Bend regions. A summed slip rate of less than 8 mm/yr is also inferred for thrust faults accommodating shortening across the Transverse Ranges. GPS-constrained, locked models suggest high slip rates for the Imperial Fault and the Coachella SAF (>35 mm/yr and >32 mm/yr) but TSE-constrained, unlocked models suggest much lower slip rates for these two faults. All models are consistent with strain transfer from the SAF Coachella segment to the Eastern California Shear Zone, bypassing the SAF, but the GPS-constrained models suggest this far more strongly. This would seem to support variations in rates of fault slip and regional deformation patterns over millennial timescales, but the comparable Mojave SAF slip rates for both model classes is inconsistent with this hypothesis. Assuming a “propeller” geometry for the SAF (Fuis et al., 2012, 2017) does not alter inferred slip rates. Correcting the GPS velocity field for seismic cycle effects associated with large earthquakes on the 1857 SAF rupture segment increases the inferred slip rate of the Mojave South segment of the San Andreas fault by up to about 5 mm/yr. “Off-fault” deformation accounts for 38% of the total moment accumulation in the preferred GPS-constrained model, broadly consistent with previous estimates.