We interpret the interseismic crustal velocity field of the central Andes using a simple three-plate model in which the Andean mountain belt is treated as a rigid microplate located between the Nazca and South American (SoAm) plates. We assume that the Euler vectors associated with these plates are strictly coaxial and that the surface velocity field can be decomposed into an elastic loading field (driven by locking of the main plate boundary) and a smaller contribution associated with back arc convergence. We obtain our best fit to the geodetic velocities if the main plate boundary is fully (100%) locked between depths of ~10-50 km and ~8.5% of Nazca-SoAm plate convergence is achieved in the back arc (by underthrusting of the Brazilian Shield beneath the Subandean zone).