Geophysical and tectonic evidence suggests that the principal vertical axis bending of the Bolivian Orocline is a late Cenozoic event related to subduction and Andean uplift. To further test this hypothesis, paleomagnetic data are presented from Miocene basins within the Altiplano and Cordillera Oriental of Bolivia, located in the presumed southern limb of the orocline. To factor out potential VGP data, samples were collected from sedimentary sections spanning at least 0.5 m.y. New paleomagnetic data are presented for two sites, i.e., Cerdas (21¿S latitude, 16--15 Ma) and Quehua (20¿S latitude, 13--7 Ma). In conjunction with previously published paleomagnetic results from middle to late Tertiary rocks from Bolivia and Peru spanning some 16¿ of latitude (6--22¿S), a consistent pattern of oroclinal development is as follows: significant counterclockwise rotation (about 15--20¿) in the northern limb, negligible rotation in the axis (at the latitude of the Arica Deflection), and significant clockwise rotation (about 15--20¿) in the southern limb. Although these paleomagnetic data could represent independent, mesoscale (≥102 to ≤104 km2) block rotations, the preferred interpretation is that they confirm the hypothesis that a significant portion (about 30--40¿) of the oroclinal bending occurred during the late Cenozoic. Using the unit mean paleomagnetic declinations in the well-dated sections, minimum rates of oroclinal bending during the late Cenozoic are calculated to range from 0.6¿/m.y. to 1.7¿/m.y. The remaining 15--20¿ of observed curvature of the orocline probably resulted from an earlier phase of Andean deformation or may represent the shape of the ancestral Andes. ¿ American Geophysical Union 1995